I have occasionally scrounged parts from discarded circuit boards. I recently was given about four audio mixer boards from a large TV production console, through a friend. This was not only a great source of pots and switches, but also had a bonanza of opamps and other associated chips.
I've traditionally extracted parts like this using a propane torch, applied to the solder side while the PCB is held in a vice on the picnic table.
You must do this outside, since it is not only very smelly, but the fumes are very bad for your health. It probably is a good idea to wear a mask. Be prepared to be stopped by your spouse-- it will attach to your clothes, and you'll need a shower afterwards also.
The propane torch approach works very well at getting parts out, since they'll often just drop out once the solder is melted. However, many parts are hooked in and will require some pulling. Chips usually pull out well, and I have a special little IC puller that hooks under each end of the chip. All of this works very well, except for the smell(!) It is of course, destructive to the PCB.
To avoid all the spousal hassle associated with the smelly process above, I tried out a different process with the remaining pair of boards (I got through the first two with the propane torch).
In the 2nd process, I used a heat gun on high. This requires considerably more patience, since it takes longer to heat up a chip (or a pot) so that all the solder melts. In fact, you may need to do a little gentle rocking etc. to help it come out.
This heat gun process however, does heat the chips considerably
more, since the heat is applied longer until the solder begins to melt. Again, you should do this outside, or at least in a well ventilated place like a garage.
I got a lot of good parts extracted this way. What I didn't know until this morning, was whether or not the extra hot chip extractions were successful or not. I managed to test several TL072, NE5523, NE5534 and a pair of LF412 opamps and happy to report that they all seemed to test ok.
To test them, I bread boarded them in unity buffer configuration, for each opamp. I supplied a signal into the non-inverting input and scoped the output to see if it looked like a good sine wave (that was the input). Someday, I want to design a generalized opamp tester along these lines and add some other fancy tests so that they can be tested without a scope.
Anyway, the point of this post is simply to say that the heat gun extraction of opamps appears to be a successful approach. The chips do get hot, but they seem to survive it ok. The trick is simply to get in there, get them out, and let them cool off as best as you can manage it.
I also found that by extracting chips in neighbouring groups is helpful, because they already have the starting heat they need (it saves time, and reduces the stress on the chips being extracted).
I've attached a picture of the dining room table as I was sorting out the parts that were extracted (while the wifey was away).
