The format starts fine, then stops - 47 percent, every single time, on every attempt. A format that freezes at the same point is not a software mood; it is the drive reporting, as precisely as it can, where its memory has died. Here is how to confirm it instead of guessing.
Random failures have random causes, but a repeatable failure is a gift: it is reproducible evidence. When a format stalls at the same percentage, fails on the same files, or dies at the same point in a copy, the drive is hitting the same damaged region of memory every time. Before condemning it, you rule out the cheap suspects - port, power, PC - and then run the one test that gives a verdict.
If the format completes on another port or PC, your drive just won a reprieve and the fault is in the first setup. If it stalls at the same place everywhere, the evidence now points at the drive itself.
A quick format only rewrites the file index - it barely touches the memory, so it can succeed on a drive full of bad cells. A full format writes across the whole surface, which is why it stalls where the damage is. This is also why a drive can quick-format happily and still corrupt your files later: the bad region was never visited. A successful quick format is not a health certificate.
The verdict test is the same one used against fake drives: H2testw fills the entire capacity with test data and reads it all back. Three outcomes are possible. It completes clean - the drive's memory is sound and your stall had another cause. It reports errors in a region - those cells are dead, permanently. It cannot even finish writing - the controller or memory has failed outright. There is no fourth outcome where a different tool disagrees; the memory either holds data or it does not.
During copies or formats, Windows may instead report: The request could not be performed because of an I/O device error. That means the PC asked the drive to read or write and the drive could not complete the operation at all. After the port and PC checks above, treat this message on a pendrive as a hardware verdict - the electronics or the memory are failing, not the file system.
Do not run repeated full formats hoping one will push through - each attempt stresses the same dying cells and changes nothing. Do not trust so-called low-level format tools with a revival; they issue the same writes to the same dead memory. Rescue any files the drive still lets you read, claim the warranty if it exists, and replace the drive. A pendrive is one of the cheapest components you own - the data you were about to trust it with is not.
See the format-failure family demonstrated in the video, including the diskpart route and when to stop trying.
▶️ Watch Full VideoA stall point can shift slightly when the damaged area is large or the controller's condition is worsening - the drive retries and remaps as it goes. A wandering stall point is still the same story: failing memory, just less tidy about it.
Technically you can partition only the healthy region, but dying flash rarely stops dying - the bad area grows, and it grows into your files without warning. For throwaway transfers it is your choice; for anything that matters, no.
A single freeze that never repeats, followed by a clean full-drive fill test, is usually a port, power or driver hiccup. Verify with the fill test once; if it passes completely, use the drive - and keep backups as you should with any pendrive.