iMac Fusion Drive Data Recovery - How We Rescued a Customer's Data From a Dying Apple SSD With Dry Ice

The failure we see over and over

On these Intel iMacs, the blade SSD carries a large Samsung controller chip, and that chip runs hot. We have a drawer full of these Apple blades, 128 GB and 256 GB, pulled from machines that came in dead. The pattern is almost always the same. The controller overheats, the drive mounts for a few seconds, and then it disconnects. Every time it powers back up it gets a little worse. The controller is also encrypted, so once it truly fails, that data is usually gone for good. So when your iMac is dropping the drive or throwing disk errors, do not keep restarting it and hoping. Every heat cycle is spending part of your last chance.

Step 1 - Get to the drives

This is not a slide-out-a-tray job. On the 27 inch iMac the storage lives behind the screen and the logic board. We removed the display, pulled the entire logic board, and separated the dual drive system so we could work on the SSD directly. The hard drive half stays in the picture too, because a Fusion volume wants to see both members before it will present itself.

Step 2 - Mount the blade in the open

We put the Apple blade into an OWC external enclosure, but left it open and uncovered so the controller was exposed and we could reach it. Closed up, we would have no way to fight the heat, and the drive would just do the same overheat-and-drop routine on USB that it did inside the iMac.

Step 3 - The unorthodox part: dry ice on the controller

Here is where it got creative. Every time I tried to pull the data, the transfer would stall out at a gigabyte or less, and the controller would get so hot I could not even touch it. So I stopped. Instead of fighting the drive, I decided to fight the heat that was killing it, and cool the controller directly. We took dry ice, sealed it in a plastic bag, and carefully held it right on top of the controller chip with a tool while the drive powered up. It worked. And here is the wild part: the controller was fighting so hard that the dry ice never even made it cold. It kept the chip merely warm instead of scorching, and that was the whole difference. Warm, the controller stayed on the bus. Scorching, it fell off. The dry ice would melt through in about twenty minutes, so it was a race, but it gave us a stable window we simply did not have before.

Cooling electronics this way is not brand new. Extreme overclockers have used dry ice and liquid nitrogen for years to push CPUs and GPUs past their limits, and I used to be one of those people. But that trick is about squeezing more speed out of a healthy chip. Using it to keep a failing controller barely alive, just long enough to rescue someone's files, is a different game, and it is not something I have seen anyone do.

Step 4 - Image everything to a healthy drive

The second the drive was stable, we started cloning. You never work off a dying drive, you copy it to a healthy one and work from the copy. We imaged the failing blade onto our own target, a 2 TB drive in a Thunderbolt 4 and USB-C external enclosure. Fast as that is, it is never the bottleneck on a job like this. A recovery runs at the speed mercy of the drive you are pulling from, not the drive you are writing to. So even on Thunderbolt 4, this crept along at whatever the dying Apple SSD would give us, and we fed the controller more dry ice every time the batch ran out. All in, this recovery took about six hours of babysitting the drive and the ice. When it was done, the clone reported finished at about 1.13 TB. The tool had first estimated 1.37 TB across 1,790,729 files, and the gap was almost certainly system files or a quirk in how it counted, because the copy completed and reported the recovery as done, off a drive that moments earlier could not stay connected long enough to move a single gigabyte. The proof is what matters: the customer went through the recovered drive himself and confirmed every one of his files was there. He could not have been happier. This method worked so well that it is now part of how we approach these overheating Apple controllers.

Step 5 - We left it better than we found it

Recovering the data was the job, but we did not stop at handing back a dead drive. With the files safe, we replaced the failing 256 GB Apple blade with a new OWC Aura X2 2 TB SSD, which is actually faster than the Apple original, installed it in the iMac, and loaded their recovered data right onto it so they could pick up exactly where they left off. We also reinstalled the customer's 2 TB SATA SSD since it was perfectly healthy, but we did not fuse the two back together. Fusion is exactly what turned one dying SSD into a whole-machine emergency, so we set them up as two separate, independent drives: the fast new OWC Aura X2 as the boot drive and the 2 TB SATA as its own storage. Now if one ever fails, it can never take the other down with it. They left with a fresh, fast boot drive and 4 TB of storage total in that iMac, more than they walked in with. The customer also made a smart call and bought the Mac Space branded Thunderbolt 4 external we recovered onto, with a full copy of their data still on it, so for the first time they have a real backup sitting outside the machine. A scare like this is the best backup reminder there is.

Logical recovery (the drive is fine, the data is not)

Accidental deletion, a corrupted volume, a failed macOS update, a reformatted disk. The hardware is healthy, so the job is software: scan the drive and rebuild the file structure. This is the most recoverable situation, as long as nothing has overwritten the data. The single most important thing you can do here is stop using the drive the moment you realize files are missing.

Physical and controller failure (the drive itself is dying)

This is the iMac above. The drive is physically failing, whether it is a spinning hard drive with bad heads or an SSD with an overheating controller. You cannot fix this in software. The work is to stabilize the hardware just long enough to make a full copy, whatever that takes, cooling a controller with dry ice included. Once we have a clean image on a healthy drive, the recovery happens off the copy where the failing drive can never make things worse.

Image first, always

The rule that saves data is simple: never work on the original failing drive. Every read you take from a dying drive is a risk, so you take that risk exactly once, to make a full sector-by-sector image onto a known-good target. From then on, everything happens on the copy, where the failing drive can never make things worse.

What NOT to do

Do not keep restarting a Mac that is dropping its drive. Do not run repair utilities over and over on a physically failing disk. Do not let it power cycle all night hoping it fixes itself. Each of those spends part of the data you are trying to save. If a drive is clicking, overheating, or appearing and disappearing, power it down and get it to someone who images before they touch anything.