2
\$\begingroup\$

For a long time now, I had a fear, that when you solder a SMD IC to a PCB, that you won't be able to desolder and resolder it again without destroying it. Recently, I did just that, but it didnt damage the chip, so I feel a bit better. I used a hot air gun at my college. I did it because not all pins were soldered proper before.

Q:

What is the rule of thumb, how many times can I desolder and again solder the chip, to save the cost on testing?

Can it be infinite or will it start producing magic smoke after a while?

\$\endgroup\$
  • \$\begingroup\$ Mass produced products often go through the reflow oven twice. I would say you should be able to rework most stuff around 3 or 4 times. \$\endgroup\$ – mkeith Jun 30 '18 at 6:26
  • \$\begingroup\$ 3 or 4 times should be OK if you are careful. The PCB itself is also an issue, assuming you are putting it back into the same place. \$\endgroup\$ – dmb Jun 30 '18 at 7:06
5
\$\begingroup\$

The rule of thumb is 'assume a chip you've desoldered from a board is damaged, until you've proved otherwise'. In other words, don't solder an SMD IC down to test it.

There's a significant difference between automated SMD soldering, in a temperature controlled oven with a calibrated warm-up and cool-down profile, and eye-balling the melting of solder with a hot-air gun. With the latter, it's so easy to get it a few 10s of degrees too hot, or dwell there for a few seconds too long.

The junctions made in an IC, often by thermal diffusion, can also be unmade by over-diffusion. The damage an IC gets from over-temperature is cummulative. The amount of damage an IC gets is also exponentially related to the temperature. It's impossible to give you a figure for how many times you can heat it to above its (often recommended) 150C limit, without being very precise about the temperature.

Whether it 'works' is a poor indicator of how much damage you've done after one solder/desolder cycle. If you instead measure the input leakage current, power supply quiescent current, noise level, gains before and after the thermal abuse, then you'll have an idea of how much shift in parameters that cycle caused, and how many more cycles it can tolerate before the parameter shifts amount to 'not working'.

\$\endgroup\$
  • \$\begingroup\$ Is the 150 C limit not generally an operation limit (as silicon PN junctions kinda stop working around that temperature)? \$\endgroup\$ – Joren Vaes Jun 30 '18 at 6:41
  • 1
    \$\begingroup\$ for certain meanings of 'kinda', yes. However, avoid using 'not' in these sort of questions, as sometimes it can mean 'not'. There are several upper limits to opersting temperature. One is bias shifts due to VBEs and suchlike changing. The other is long term damage due to diffusion, for which 150C is often given as a storage temperautre limit to be 'a long way' from the temperature at which damaging diffusion can accumuklate in the lifetime of the device. To this end, manufactueres tend to make both temperatures equal. \$\endgroup\$ – Neil_UK Jun 30 '18 at 6:48
4
\$\begingroup\$

This is about a socketed part instead of soldered, but here is one data point from the datasheet of the famous Signetics 25120 Write Only Memory:

\$\endgroup\$
  • \$\begingroup\$ number of remaining pins! lol !!! \$\endgroup\$ – abu_bua Jun 30 '18 at 16:10
  • \$\begingroup\$ Can i conclude, that if I insert the device more than 60 times, i have -2 remaining pins? \$\endgroup\$ – abu_bua Jul 3 '18 at 11:54
3
\$\begingroup\$

I think this is an interesting question.

Let us assume that you do not physically damage the chip in the process during the desoldering process. Let us also assume that this is not a device that has BGA balls mounted on it that would be lost when you desolder it, and be hard to replace.

The only thing I can imagine resulting in damage and limiting the amount of times you can solder/desolder a part is the impact of the heat on the junctions. The junctions within your chip are built with dopants that form P and N doped areas (and within those areas we still have different gradients). There are also metal-semiconductor junctions with barrier layers in between.

These dopants slowly move do to diffusion - even at room temperature. But the rate at which they move increases exponential, making it hotter makes them move faster. The only thing I can imagine limiting the amount of resoldering is the constant heating impacting this diffusion. Especially in the front-end (which is what IC manufacturers use to refer to all the interconnect structures) this might be an issue, if e.g. copper from interconnects can start breaking through barriers again and again and cause issues.

However, I think the time it will take before this becomes an issue will be quite large, especially if you are using hot-air soldering, and thus you could say that you can indeed re-solder an infinite amount of times.

Note that if you want to do this to reuse a chip during prototyping or such, there are special sockets for pretty much all packages that allow you to connect a chip without soldering (ZIF sockets). They are generally not cheap at all, but they will be a lot cheaper than say ruining a very expensive chip, or even cheaper than the time it might take to try and safely desolder the chip and put it on another board. I have seen them used for verifying silicon prototypes where you can't go out and buy more chips because they 20 dies are all that exist.

\$\endgroup\$
0
\$\begingroup\$

If you read a good datasheet carefully, you will detect something like temperature mission profile.

It list the estimated life-time until a failure occurs. E.g.

50°C ... 10^7 hours
100°C ... 10^4 hours ...etc.

That means, the supplier guarantees a life-time when the device operates max. 10^7 hours with temperature lower than 50°C, 10^4 hours < 100°C. ..

Also a soldering profile should be given, which looks like this one:

enter image description here

The underlying physics is the reaction kinematics of Arrhenius (believe it or believe it not):

$$k = A \cdot \mathrm{e}^{-\frac{E_\mathrm{A}}{R \cdot T}}\, .$$

The more often you heat up your device, the more you strain the device, until it fails.

What happens, for instance:

  • probability that atoms displace is higher for increased temperature! -> results in a change of doping profiles, change in threshold voltage of mos transistors, etc..

  • electromigration, damage of metal layer

A temperature mission profile strongly dependens on the semiconductor process (design rules, type of circuit, how it is layouted, ...).

In order to answer your question, how often you can solder/desolder a device depends therefore on the time and heat during the soldering, and on the device itself.

\$\endgroup\$

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.