0

I was trying to solder some VQFN packaged chips for the first time using a hot air gun, because that's all I have. I set the temperature to rougly < 350°C and startet the process. Sometimes it wouldn't workout perfectly from the beginning on so I had to hold it for a bit longer (up to 30-50 seconds) or make some corrections using a solder iron. A lot of these chips were not working properly. Later on I figured out how to do it faster and manged to get two of them to work.

Now I started to wonder if I really destroyed these chips by overheating them. I really don't think the silicon or any other material was damaged by this. But what about the bond wires? Is it possible that they desoldered/detached?

MazzMan
  • 101
  • 1
  • 1
    "I really don't think the silicon or any other material was damaged by this." Why? There's no good reason to think this. Reflow profiles specify a peak temperature much lower than 350C and hold it much shorter than 30-50 seconds. Have you not seen what happens when you hold a soldering iron at 350C for even 20 seconds on a chip? – DKNguyen Oct 07 '19 at 21:20
  • 6
    Bond wires are not soldered, they're welded. – Dave Tweed Oct 07 '19 at 21:22
  • 1
    It is possible to damage an IC by overheating it. But it is also possible that your first efforts at soldering were plagued by solder bridges, cold solder joints, contamination in the solder, etc. The more you do it the better you get. I doubt the bond wires actually lifted. – user57037 Oct 07 '19 at 21:33

3 Answers3

2

But what about the bond wires? Is it possible that they desoldered/detached?

The bond wires generally aren't soldered to the chip or to the package. They're pressure welded.

If the bond wires are embedded in the plastic package material (I believe they will be for most QFN package types), and you heated the plastic beyond its glass transition temperature, you could well have caused the plastic to pull the bond wires apart. But you'd likely see burning of the outer surface of the plastic before you did this.

The Photon
  • 126,425
  • 3
  • 159
  • 304
1

With some caveats, it is actually possible to test if the bond wire is connected if you remove it from the PCB. First of all take a known good IC. Then use a volt meter in diode mode. Connect the black probe to IC GND and the red wire to one of the pins you want to test. On the known good IC note the forward voltage shown in diode mode. You may also want to measure in OHM mode to see what it says.

Now, test with the exact same setup on the suspect IC. If you get a substantially different result, the IC is probably bad. If you get an open-circuit reading on the pin, then the bond wire could actually be broken. Confirm in Ohm mode.

This test is not perfect. But by and large if the wire is not broken, you will get some kind of reading with the Ohm meter either in Ohm mode or diode mode. For sure if there is any conductivity to GND, then the bond wire is likely not broken.

user57037
  • 28,915
  • 1
  • 28
  • 81
0

Now I started to wonder if I really destroyed these chips by overheating them. I really don't think the silicon or any other material was damaged by this. But what about the bond wires? Is it possible that they desoldered/detached?

No, but if you didn't follow the reflow profile, you could have damaged the chip. There is no guarantees if the reflow profile is not followed. Typically this is not the case, I've soldered VQFN's with hot air and not had issues.

If you are having issues with chip death during reflow, then make sure you do these things:

  • Follow proper ESD procedure
  • Make sure chips are not exposed to the Moisture Sensitive Level (MSL) beyond what the are rated before reflow (moisture can create steam and mechancially destroy chips, especially MEMs and optical parts)
  • Follow the reflow profile

If you do the above things, then your chips are guaranteed to work. For the people that I've consulted with that were having problems, they were usually not following MSL or ESD procedures. Once those were implemented, the problems went away.

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208