47
\$\begingroup\$

Our purchasing department requested the AT32UC3B1256 and the supplier has a few thousand in storage. However, those trays have a date code from 2009.

While shelf life is normally not much of an issue with integrated circuits, 10 years is quite a number.

I looked for some general information regarding shelf life. For example TI states:

[...] TI’s standard shelf life for packaged products is two years from the time it was manufactured to the time it is delivered by TI or a TI authorized distributor.
TI also offers extended shelf life (ESL) of certain products for up to five years of total shelf life from the time it is manufactured to the time it is delivered by TI or a TI authorized distributor Product warranty is measured from the actual shipment date, not the date of manufacture.

According to supplier, the trays are sealed. What would be the limiting factor for shelf life with respect to microcontrollers? May those parts require long tempering before assembly?

TLDR: Should we be concerned about buying a 10-year-old microcontroller?

\$\endgroup\$
12
  • 10
    \$\begingroup\$ I know moisture is a big worry. Often the trays have a moisture 'label' which changes colour. I would start with asking if they are present and if so what their status is. \$\endgroup\$
    – Oldfart
    Jan 24, 2019 at 9:23
  • 11
    \$\begingroup\$ The normal risk is moisture causing problems during reflow; the technique for dealing with these is to dry bake the chips at just above 100C for 24H. \$\endgroup\$
    – pjc50
    Jan 24, 2019 at 9:34
  • 12
    \$\begingroup\$ Also, since it is a microcontroller, it may have bugs that were fixed in later revisions. So one should probably test the code on the actual chip (same batch/date code) before using it for production. \$\endgroup\$
    – Indraneel
    Jan 24, 2019 at 9:59
  • 4
    \$\begingroup\$ last-resort only , shop for a retinning facility that will bake, with ESD handling your batch, Purchasing is trying to save a buck on a $5 part that may end up costing more \$\endgroup\$ Jan 24, 2019 at 10:13
  • 2
    \$\begingroup\$ These are parts are in stock at many distributors, the cost of re-processing oxidized , mositure ingress issues with possible lead coplanaraity handling damage, ESD damage may exceed the savings from a final sale surplus supplier. These parts have a Moisture Sensitivity Level(MSL): 3(168 Hours) THey should not be baked near or above 100'C \$\endgroup\$ Jan 24, 2019 at 10:27

5 Answers 5

49
\$\begingroup\$

Actually, I found an interesting document from TI: Component Reliability After Long Term Storage).

I will just quote some interesting parts here:

  • Each year the semiconductor industry routes a significant volume of devices to recycling sites for no reliability or quality rationale beyond the fact that those devices were stored on a warehouse shelf for two years.
  • Results show that current packaging material (mold compound and leadframe) is sufficiently robust to protect the active integrated circuits for many decades and permit standard reflow solder assembly beyond 15 years.
  • Standard packing materials (bags, desiccant, and humidity cards) are robust for a 32 month storage period that can be extended by repacking with fresh materials.
  • Packing materials designed for long term storage are effective for more than five years.

Background:

  • The origins of date code age restrictions are not well documented, but it is probable that limitations of the packing materials for moisture sensitive components and post storage solderability of SnPb or Sn finishes contributed to the concerns of customers that led to the shelf life restrictions

Device risks:

  • Device functionality and parametric performance after extended periods of shelf storage
    • No failure mechanisms have been identified that would compromise the electrical performance or circuit reliability of LTS devices.
  • Exposure to the ambient atmosphere for extended periods of time may oxidize the lead surface impacting solder wetting during assembly.
    • Aging studies have shown that NiPdAu lead finish devices pass solderability requirements beyond 8 years. Actual testing of LTS devices indicates that storage for 15 years does not compromise solderability.
  • Moisture absorbed in the epoxy matrix of devices categorized as MSL 2 to 6 could vaporize during reflow solder assembly and crack the package.
    • Devices targeted for LTS are packed in special metallized bags that are sealed with desiccant and a HIC.
  • Devices stored for extended periods may exhibit corrosion of bond pads or interconnect metallization
  • Devices that incorporate non-volatile memory may suffer data retention issues after long storage intervals.

Packing material risks:

  • The static dissipative properties of tubes or tape and reel may degrade over time resulting in potential ESD damage to LTS devices
  • Storage bags may leak, allow moisture to enter, and cause problems for MSL
  • Label adhesives may fail or the ink marking could fade making lot history or device identification difficul

Conclusion:

The shelf life of LTS devices as determined by solderability, SEM visual, SEM spectral analysis, optical microscopy, MSL performance, solderability, and decapsulation/visual is >15 years.
IC packing material shelf life is limited by moisture diffusion through the MBB. A standard MBB maintains satisfactory moisture levels for 32 months. LTS bags control moisture levels beyond 5 years.

Terms:

  • Humidity Indicator Card: (HIC) – A card printed with a moisture sensitive chemical (cobalt chloride) that changes from blue to pink in the presence of water vapor.
  • Long Term Storage: (LTS) – Storage of devices in an uncontrolled indoor environment for more than two years.
  • Moisture Barrier Bag: (MBB) – Storage bag manufactured with a flexible laminated vapor barrier film that restricts transmission of water vapor.
\$\endgroup\$
32
\$\begingroup\$

The concern is solderability.

If not stored correctly oxidation on the pins can cause problems in the solder process.

Try ordering a few to begin with and see how it goes.

Your EMS might have some tricks to deal with components that have been on the shelf too long.

\$\endgroup\$
2
  • 11
    \$\begingroup\$ This can be a major problem. As some annec-data, we had a push a while back to use up non-moving stock. We had an enormous amount of 10k 1206 res arrays, so they were designed into a new product. Would the things reflow? Would they heck! It was even hard to rework them with an aggressive flux.They ended up in the skip and we had to buy new ones. \$\endgroup\$
    – Cursorkeys
    Jan 24, 2019 at 11:18
  • 4
    \$\begingroup\$ Similarly watch the age of the PCBs themselves. Same problem if you exceed stated shelf life especially with some of the newer "green" finishes -- that one was fun since EMS didn't follow PCB vendor data and quality issues suddenly appeared. \$\endgroup\$ Jan 24, 2019 at 16:24
26
\$\begingroup\$

I would not worry too much about the chips functioning.

What I would be relatively concerned about is solderability- that’s the main reason behind the relatively short “shelf life” numbers. A more aggressive flux may help- there may be some recommended procedures used by low volume high-rel folks who may have little choice about using NOS (New Old Stock) parts. At the other extreme, I’m told one car maker has a 6-month limit on parts - reflecting the perceived risk of unreliable PCBs vs. their ability to dominate suppliers.

Also, if you are not 100% sure they were stored in a low humidity cabinet, a specified bake procedure is called for to drive moisture out of the plastic packages and prevent damage in the reflow process- it’s unnecessary for hand soldering.

\$\endgroup\$
4
  • \$\begingroup\$ What is a NOS part? Re car makers: I heard that the boards in cars need to be made of special materials to withstand the temperature shocks and gradients and the acceleration cycles (bumps and vibration) long term, which may affect solderability and put more stringent requirements otherwise. \$\endgroup\$
    – Pavel
    Jan 25, 2019 at 5:35
  • 4
    \$\begingroup\$ New old stock (NOS) parts are parts that have been stored for a relatively long time, \$\endgroup\$ Jan 25, 2019 at 6:32
  • 2
    \$\begingroup\$ Automotive requirements are more stringent than some commercial requirements, and things like some toys, but often less so than medical, aviation and space requirements. See, for example, IPC J-STD-001 and relevant inspection standards, international standards which have replaced internal proprietary standards at places like NASA. \$\endgroup\$ Jan 25, 2019 at 6:45
  • 1
    \$\begingroup\$ In some cases, it's necessary to purchase recycled parts which have been pulled from unused PCBs, used PCBs or PCBs from product which has been returned during a warranty period. In China, for example, such parts are available repackaged in tape and reel, and of course an honest supplier will disclose that they are second hand (written literally as 二手). \$\endgroup\$ Jan 25, 2019 at 6:51
18
\$\begingroup\$

In my opinion the shelf-life might be more limited by whoever wants to take the responsibility to guarantee that the ICs still work rather than that there are technical reasons for the ICs to "expire". So more a legal responsibility than a technical issue.

I think it is very likely that the ICs will just work without issue as they were stored in a proper way, in a dry environment.

If the price of these ICs was considerably less than "new" ones and your company is willing to accept that there might be a small chance that there are issues with these ICs then it can be a good deal.

If you prefer "no risk" and "guaranteed to work" then you might need to use "new" ICs.

You do not mention what the supplier states about the fact that these ICs are 10 years old, do they still guarantee 10 years lifetime for example?

\$\endgroup\$
3
  • 7
    \$\begingroup\$ This has to do with solderability, not with the functionality of the ICs as such. \$\endgroup\$
    – Lundin
    Jan 24, 2019 at 10:02
  • 2
    \$\begingroup\$ @Lundin I agree, although I did not mention "solderability" as such I see it as included in "functionality", if you can't solder it, it doesn't work right? Also if you can solder it but it does not work, you still have no functionality. \$\endgroup\$ Jan 24, 2019 at 10:03
  • \$\begingroup\$ Spot on, for the last point, jn my experience resellers (unlike distributors) will not guarantee anything beyond that it is genuine. Good ones allow you to perform due dilligence, but even with samples it is hard to tell , you can have one bad reel or tray and they've given you 5 samples off the first you would likely eat that cost. I would also request samples of each lot if there is more than one. \$\endgroup\$
    – crasic
    Jan 25, 2019 at 9:07
11
\$\begingroup\$

Unless there is some immediate need, Find a better distributor. There is no reason to buy 10 Y.O date codes of an active part. It is more trouble than it's worth.

That being said, properly stored and cared for IC components will be fine to mount after 10 years.

The key here is properly stored. With older parts that pass through several hands there is a real possibility that someone screwed up. Unless you have proof positive of their condition and storage (usually certificate from the reseller), it is more trouble than it's worth for production use.


It looks like you are doing spot buy from a re-seller, likely not officially authorized distributor of Microchip who would not hold old lots like this. If the existing distributors do not suffice, Contact microchip directly and ask for assistance sourcing QTY 1500 of this device, they are friendly and will provide you a rep to talk to that will help. Email them!

Most of the other answers are based on the technical feasibility of these chips being OK, which to me is not relevant. Those guidelines apply when you can guarantee good storage practices going back to when the parts were first sold. You have no idea how many hands it has passed through or how they stored it unless distributor is willing to certify to that. I like @Bimpelrekkie answer because he frames it as the commercial issue that it is.

Lacking that, this is a bad idea unless you are prepared to test, your CM/EMS may or may not be able to help you with this. It is a massive headache and not worth it (IMO) unless you don't care or are desperate.

Moisture dots can be changed, leads can be cleaned or poorly replated in a non-obvious fashion, parts can be re-reeled, silkscreens re-printed, and labels restickered.

Things I have personally seen that have gone wrong in this situation when using second market parts.

  1. Counterfeit Part

  2. Desoldered/re-trayed/re-reeled parts

  3. Captive Moisture leading to part failure during reflow or soon during operation

  4. Lead corrosion due to poor storage.

  5. Old stepping or revision with subtle issues that you did not qualify in the design.

    (Always Read the Chip Errata!, best one I found so far was a memory IC with erata that read "ECC on this ECC part never worked, fixed in rev B.", discovered due to high corruption rate when using Rev A part due to shortage)

What I have done in this situation to mitigate when there was simply no way around (Obsolete part, global shortage, cash strapped startup)

  1. Always Get Samples
  2. Read the erata
  3. Contract with distributor on guaranteed refund in case of non-usable parts
  4. Give to 3rd party lab to examine carefully
  5. Have CM/EMS perform extended bake-out and solder ability tests
  6. Decapsulate and image IC compared to new lot.
  7. Carefuly test first PCBA articles using suspect lot
\$\endgroup\$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.