Frozen Bytes

One trick I learned early in my computer career served me well on numerous occasions: freezing hard drives.

While this certainly did not work every time, for many issues with the older platter-style hard drives, simply putting it in the freezer for anywhere from 3 to 24 hours could temporarily resurrect a drive. Common ailments this could treat included the dreaded "click of death", the "unrecognized media" errors and spontaneous system faults (which typically occur leading up to the other two).

Often what I found worked best was freezing the drive, then running the Steve Gibson program, Spinrite, on it to detect problem sectors, copy the data to functioning sectors, and then flagging the problem sectors so they were not written to again. At this point you could either put the drive back into service (not recommended), or if it continued to exhibit problems, copy the data to a new drive and dispose of the old one. I found using either the Windows safe mode or a bootable BartPE disk would work best for copying the data. Since it was mounted using just a lightweight operating system, fewer resources were consumed, and it was least impactful on the drive itself since the operating system would not run processes to index all of the files.

One drive I dealt with was particularly problematic.

The drive came out of the computer belonging to the chair of the Management Information Systems department in the College of Business. And of course, there were no backups taken of any of the data stored locally (which unfortunately was most of the data they needed). Usually you could tell pretty quick if the freezing technique was going to work on a drive - you freeze it and either works or it doesn't. In the case of this drive though, it would start to work and then it would stop. At first I suspected there was a sector that was so corrupt whenever it would try to be read, the whole thing would crash. But after the second or third time freezing it and then starting a copy, I noticed that the copy would error out in different places. One time it might error out after 11%. The next time it might error out at 27%. Something didn't add up. The really unfortunate part was that Spinrite wanted nothing to do with the drive at all. It would start up and immediately hang no matter what.

Since I could get some of the data to copy to different points, I wondered if whatever the freezing technique was overcoming would arise again as the drive warmed back up? To test this theory, I devised a way to keep the drive frozen while retrieving data from it. Remembering how ice and salt interact from my physics classes, I thought, maybe if I could create an environment colder than ice, I could lengthen the time until the drive inevitably crashed again.

I gathered some supplies: an old 5.25" floppy disk storage container, a static resistant bag, an additional bag (make doubly sure not to get the drive wet), some ice, and some table salt. I put the drive in the static bag, the static bag in the other bag, put those in the floppy disk storage container partially filled with ice, put more ice on top of the bags, connected the hard drive to the computer I used to copy files, and then salted the ice (lowering the freezing point, thus maintaining a lower temperature of the ice bath). By doing this, I was able to retrieve far more data from the drive - almost 85% of the data was recoverable. While I was not able to recover everything, there is no way I could have preserved as much of their data as I did, had I not discovered this solution.