I once had a hard drive fail this way. It was a 60GB SATA disk from around 2006. I powered down the machine to drop in a couple of new disks and gave the machine a much needed dusting.

When it restarted it had the click of death. The freezer trick worked long enough for me to grab the few kb of data I really needed from that disk.

I wrote about it here: https://miscdotgeek.com/adventures-in-hard-drives/

If you are powering off a very old system that spends most of its life powered on, don't let the hard drives or power supply cool down. Do your best to keep your work as brief as possible and get the thing turned back on before it assumes room temperature. These components tend to run the hottest, you can have cold solder joints contract and stop making contact when they cool down.

Also check any socketed chips and make sure they are fully seated. Chip creep happens when frequently thermal-cycled boards cause the IC's to slowly lift themselves out of their own sockets.

That's because you need to sneak behind them with a Hiren's Boot CD before the hard drive has time to kill itself.

Most likely because the heads try to read a few back blocks that weren't accessed for years.

You can do block checks regularly, which takes a looong time on high capacity drives (or is impossible on newer ones due to firmware restrictions).