How NAND Flash Cells Wear Out
Every NAND flash cell stores data by trapping electrons in a floating gate (planar NAND) or a charge trap layer (3D NAND). Writing data requires a high voltage pulse to push electrons through a thin oxide barrier. Erasing requires the opposite voltage to pull them back out. Each program/erase (P/E) cycle degrades this oxide layer. After enough cycles, the oxide becomes too damaged to reliably hold a charge, and the cell can no longer distinguish between a stored 1 and a stored 0.
Modern consumer SSDs use TLC (3 bits per cell) or QLC (4 bits per cell) NAND, which stores more data per cell but at the cost of tighter voltage margins. A TLC cell must distinguish between 8 voltage levels; QLC must distinguish between 16. As the oxide degrades, these voltage windows drift and overlap. The SSD's ECC (error correction code) engine compensates up to a point. When the uncorrectable bit error rate (UBER) exceeds the ECC capability, the controller marks the block as bad and retires it.
Once the pool of spare blocks in the overprovisioned area is exhausted, the controller has no room left to relocate data away from failing cells. This is when the drive transitions to read-only mode: the controller stops accepting writes to prevent corrupting existing data with unreliable cells.
NAND Endurance by Cell Type
The number of bits stored per cell directly affects endurance. More bits means tighter voltage margins and faster oxide degradation.
| NAND Type | Bits/Cell | Voltage Levels | Typical P/E Cycles | Common Use |
|---|---|---|---|---|
| SLC | 1 | 2 | 50,000 - 100,000 | Enterprise cache, industrial |
| MLC | 2 | 4 | 3,000 - 10,000 | Enterprise, prosumer |
| TLC | 3 | 8 | 1,000 - 3,000 | Consumer SSDs (mainstream) |
| QLC | 4 | 16 | 500 - 1,000 | Budget SSDs, read-heavy storage |
These are typical rated endurance values. Actual lifespan depends on workload, operating temperature, controller firmware, and overprovisioning strategy. Higher temperatures accelerate oxide degradation.
Symptoms of NAND Wear Failure
NAND wear rarely causes a sudden crash. The degradation is gradual, and the symptoms escalate over time as more blocks are retired.
Stage 1: SMART Warnings
SMART attributes begin reporting degradation. The Media Wearout Indicator (Intel SSDs, attribute 233) or Wear Leveling Count (Samsung, attribute 177) drops below the manufacturer's threshold. CrystalDiskInfo or smartmontools flags the drive as "Caution." Write performance may degrade as the controller relocates data away from weak blocks. The drive still functions normally for reads and writes.
Stage 2: Read-Only / Write-Protected Mode
The controller disables write operations. The drive appears in the operating system and files are readable, but any attempt to write, delete, or modify files fails. Windows may report "The disk is write-protected." Linux shows "Read-only file system" in dmesg. This is the controller's last-resort data preservation mechanism. The drive is telling you to copy your data off before it can no longer guarantee read integrity.
Stage 3: Sector Read Failures / Controller Lockout
Cell degradation has progressed past the ECC threshold on multiple blocks. Read operations return I/O errors on specific files or directories. The operating system may freeze when accessing damaged regions. In severe cases, the controller enters a fault state and the drive stops responding entirely, appearing as 0 bytes or disappearing from BIOS. At this stage, PC-3000 SSD is required to bypass the controller and image whatever data remains accessible on the NAND.
Write Amplification and Why It Accelerates Wear
NAND flash cannot overwrite data in place. To modify a single page (4-16 KB), the controller must read the entire block (typically 256 pages, or 1-4 MB), erase the whole block, and write the updated data back. This means a single 4 KB user write can trigger a full block erase and rewrite, consuming one P/E cycle on every cell in that block.
The ratio of actual NAND writes to host writes is the write amplification factor (WAF). A WAF of 3x means the SSD writes 3 bytes to NAND for every 1 byte the host sends. Garbage collection, wear leveling, and metadata updates all contribute to WAF. Random small writes produce the highest amplification because they fragment data across many blocks, forcing more full-block erases.
TRIM support reduces write amplification by letting the controller know which blocks are no longer in use, so they can be erased proactively during idle garbage collection rather than during active writes. An SSD without TRIM enabled (common when used with older RAID controllers or in external USB enclosures) accumulates higher WAF and wears faster.
How TRIM and Garbage Collection Behave on a Near-Worn Drive
TRIM is a logical deallocate command, not a physical erase. When you delete a file or empty the recycle bin, the operating system sends TRIM to tell the controller which logical block addresses are no longer in use. The controller unmaps those addresses from the flash translation layer, and any later read of them returns deterministic zeros (DZAT).
The physical NAND cells are not erased the instant TRIM runs. Garbage collection erases the underlying blocks asynchronously, later, to reclaim them for new writes. The mechanics of this are covered on our SSD wear leveling reference page.
On a drive near end of life, this ordinary housekeeping turns hostile. Wear leveling and garbage collection both relocate live data between blocks to even out P/E cycle usage. Every relocation spends a fresh program/erase cycle, and on a worn drive those cycles run through cells that are already past their margin. Background activity that extends the life of a healthy SSD accelerates the failure of a worn one, because it keeps forcing writes into degraded oxide.
There is a second, sharper risk. Until garbage collection actually erases a block, a recently deleted file may still have a readable copy sitting in NAND. A TRIM-triggered erase on that block removes the only remaining copy of data the controller has already unmapped. Once the controller has unmapped a block, it stops returning that block's data over SATA or NVMe, and no lab can pull it back through the controller. We describe the cell-level side of this on our NAND degradation page.
This is why imaging promptly matters. The longer a worn drive stays powered with a live filesystem, the more idle-time garbage collection runs, and the more relocations and erases churn through marginal cells. Pulling the drive and imaging it in a controlled read-only pass with PC-3000 SSD captures the current state before the controller's own maintenance erodes it further.
What Not to Do With a Worn SSD
A wear-exhausted SSD is in a fragile state. The following actions can push it from recoverable to unrecoverable.
- ✗Do not run a secure erase or format. Secure erase commands force P/E cycles on already degraded cells. This can push cells past the point where the ECC engine can reconstruct the data, turning a read-only drive into an unreadable one.
- ✗Do not flash the firmware. Firmware updates write to the NAND area reserved for the Flash Translation Layer (FTL) and firmware modules. On a wear-exhausted drive, these writes can corrupt the FTL mapping, destroying the logical-to-physical address table that maps your files to specific NAND pages.
- ✗Do not repeatedly power cycle the drive. Each power-on forces the controller to read its firmware from NAND and rebuild the FTL in DRAM. On worn NAND, this read operation itself introduces bit errors. Charge stored in degraded cells also leaks faster when the drive is powered off (data retention loss), so alternating between powered and unpowered states accelerates degradation.
- ✗Do not run data recovery software that writes to the source drive. Some recovery tools write scan results, logs, or temp files to the source drive by default. On a wear-exhausted SSD, any write operation risks triggering garbage collection on degraded blocks. If the SSD is still writable, use a tool that only reads; better yet, clone the drive first and scan the clone.
How We Recover Data From Worn SSDs
Recovery from a wear-exhausted SSD is primarily an imaging challenge. The data is still on the NAND, but the cells holding it are unreliable. The goal is to read every accessible sector before further degradation makes them unreadable.
1. SMART and Controller State Assessment
We read the full SMART attribute table to determine wear level, reallocated sector count, program fail count, and erase fail count. We check whether the controller is in normal mode, read-only mode, or a fault state. This determines the recovery path: if the controller is functional and the drive is in read-only mode, we can image through normal SATA/NVMe channels. If the controller is locked, we use PC-3000 SSD.
2. PC-3000 SSD Imaging
For drives that are still readable but degraded, PC-3000 SSD performs a sector-by-sector image with configurable retry counts and timeout settings. Sectors that return read errors on the first pass get multiple retry attempts with adjusted read voltages. The tool logs every bad sector for later analysis. For drives in a controller fault state, the PC-3000 can inject volatile microcode through the controller's Technological Mode interface, temporarily restoring operation without writing to NAND.
3. Bad Block Mapping and Data Reconstruction
After imaging, the PC-3000 generates a map of readable vs. unreadable blocks. We cross-reference this against the FTL to determine which files are affected by the bad blocks. In many NAND wear cases, the majority of the drive's data is intact; the damage is concentrated in heavily-written areas like the operating system partition, swap files, and database journals while user documents, photos, and infrequently-modified files remain fully readable.
Firmware Fault vs Physical Cell Exhaustion
A worn SSD that stops responding can be failing in two different ways, and they take two different recovery paths. The first question in the lab is whether the controller is alive and confused, or alive and reading dead cells. Sorting that out before touching the drive decides whether the next step is firmware re-initialization or sector imaging.
Firmware Panic / ROM Mode: A Logical Fault
Sometimes the cells holding the data are fine, but the controller cannot load the firmware overlays it needs to present the drive. On a worn drive, the system area that stores those firmware modules can itself accumulate read errors, and the controller drops into a minimal boot state instead of full operation. The drive then shows up as 0 bytes, a raw model string, or a generic identifier, even though user data is untouched. This is a firmware-level fault, not cell death.
When the controller is in this state, PC-3000 SSD can re-initialize it in technological mode, loading volatile microcode into the controller to restore a working state without writing to NAND. The original firmware modules are read, the translation tables are rebuilt, and the drive presents its user area again. Our SSD firmware panic and ROM mode page covers this failure class in depth.
Physical Cell Exhaustion: ECC-Level Read Failures
The other case is a controller that boots and answers normally, but individual reads fail ECC because the cells underneath have lost their voltage margin. There is no firmware trick for this. Recovery is sector-by-sector imaging with PC-3000 SSD using elevated read-retry counts and read-voltage threshold (Vref) adjustment, which shifts the reference voltages the controller uses to interpret a cell so a marginal charge level still resolves to the correct bit. Each pass logs which sectors came back and which did not, and damaged regions get repeated retries before the surrounding cells degrade further.
Why the Distinction Drives Pricing
A firmware fault on otherwise healthy NAND and physical cell exhaustion sit in different effort bands, so the triage result maps onto our published SSD recovery tiers. We run the assessment first, then quote. Free evaluation, firm quote before any work begins, and no data means no charge.
SSD Models Known for Early Wear Issues
Some SSD models are more susceptible to premature NAND wear due to controller firmware bugs, insufficient overprovisioning, or early-generation QLC NAND.
Samsung 990 Pro: Firmware-Driven Wear Anomaly
Multiple Samsung 990 Pro users reported rapid SMART health drops (from 99% to 60% within weeks) caused by a firmware bug that triggered excessive internal writes. Samsung released a firmware update to address the issue. Drives affected before the update may have consumed a disproportionate amount of their NAND endurance. See our Samsung 990 Pro firmware degradation page for details on this specific issue.
Budget QLC Drives: Limited Endurance by Design
Drives like the Intel 660p, Crucial P1, and early Sabrent Rocket Q use QLC NAND with P/E cycle ratings around 500-1,000 cycles. Under sustained write workloads (surveillance cameras, database servers, scratch disks), these drives can exhaust their TBW rating in 1-3 years. Their SLC write caches mask the underlying QLC performance, but once the cache fills during sustained writes, performance drops and wear accelerates.
DRAM-less SSDs: Higher Write Amplification
Budget SSDs without a dedicated DRAM cache (using HMB or no external buffer) store their FTL mapping tables directly on the NAND. Every FTL update requires a NAND write, increasing write amplification. Controllers like the Phison S11 and Silicon Motion SM2259XT are common DRAM-less designs found in budget drives such as the PNY CS900 and some revisions of the Kingston A400. These drives wear faster than DRAM-equipped models under the same workload.
SMART Attributes That Signal Wear Exhaustion
Not all SSD manufacturers use the same SMART attribute IDs. These are the key wear-related attributes across major controller families.
| Attribute | ID | Used By | What It Means |
|---|---|---|---|
| Media Wearout Indicator | 233 | Intel | Percentage of rated NAND endurance remaining. 0% = rated lifespan consumed. |
| Wear Leveling Count | 177 | Samsung | Counts down from 100 to 0 as P/E cycles are consumed. Raw value = total P/E cycles used. |
| SSD Life Left | 231 | Vendor-specific (varies by controller) | Percentage of remaining drive life. Some controllers trigger read-only mode at 0%. |
| Percentage Lifetime Used | 202 | Micron / Crucial-class controllers | Endurance-remaining indicator. The raw value counts up toward an exhaustion threshold as rated P/E cycles are consumed. |
| Percentage Used | NVMe Log | All NVMe drives | NVMe spec-defined. Counts up from 0%. 100% = rated endurance consumed. Can exceed 100%. |
| Available Spare | NVMe Log | All NVMe drives | Percentage of spare blocks remaining. Below threshold triggers controller warnings. |
How to Read Wear Leveling and Lifetime SMART Values
SMART wear attributes carry two numbers, and the one that matters is usually the normalized value, not the raw value. The normalized value is a 0-to-255 scale (most wear attributes start at 100) that the controller maps from the underlying raw counter. The drive ships an attribute-specific threshold, and the rule is simple: the closer the normalized value drops toward that threshold, the closer the drive is to the point where the controller restricts writes.
Attribute 177 Wear Leveling Count works this way on Samsung and several other controllers. The normalized value counts down from 100 as the average program/erase count across blocks climbs, and the raw value reports the worst-case or average erase count behind it.
A normalized 177 sitting just above the vendor threshold means the average block has burned most of its rated cycles, not that a single block failed. Treat a 177 approaching threshold as a back-up-now signal rather than a fixed point at which data is lost.
Attribute 202 Percentage Lifetime Used, used by Micron and Crucial-class controllers, reads in the opposite direction. Its raw value counts up toward an exhaustion threshold as endurance is consumed, the way the NVMe Percentage Used field does. Because vendors normalize and threshold these counters differently, there is no single raw cutoff that applies across every drive.
Read the attribute by its semantics: a normalized value at or near the vendor threshold, or a lifetime-used counter near its rated ceiling, both say the same thing, which is that overprovisioning is nearly spent and read-only mode can arrive at any write.
NAND Wear and Data Retention
A worn NAND cell loses its stored charge faster than a fresh one. The JEDEC standard (JESD218) specifies that consumer SSDs must retain data for 1 year at 30°C when powered off. This specification assumes cells at their rated P/E cycle limit. Cells that have exceeded their rating retain data for shorter periods.
Time-sensitive: If your SSD is in read-only mode due to NAND wear, leaving it powered off for extended periods (months) allows the degraded cells to lose their charge. Data that is readable today may become unreadable after prolonged unpowered storage. If you cannot back up the data yourself because the drive is read-only, send it for professional imaging promptly.
Temperature matters. NAND charge leakage accelerates at higher storage temperatures. A worn SSD stored in a hot attic or car trunk degrades faster than one kept at room temperature. Powered-on SSDs retain data better than powered-off ones because the controller performs background ECC scrubbing and data refresh operations that detect and correct bit errors before they become uncorrectable.
Thermal Data Retention and the JESD218 Application Classes
JEDEC's JESD218 standard sets the powered-off data retention a drive must hold when its cells sit at the rated P/E cycle limit, and it sets that requirement differently for two application classes. The consumer/client class is specified to retain data for 1 year at 30°C powered off.
The enterprise/data-center class is held to a shorter retention requirement measured at a higher temperature. Those are the two endpoints of the same standard: a client drive trades retention time at a lower temperature, an enterprise drive accepts less retention but is qualified warmer for always-on duty.
Both of those figures describe a cell at its rating, not a cell past it. A NAND cell that has exceeded its rated P/E cycles holds charge for a shorter span than the spec minimum, because the worn oxide leaks the trapped charge faster. The standard tells you the floor for a healthy worn-to-rating cell; a drive that is already in read-only mode from wear is, by definition, operating below that floor.
Temperature pushes the trend the wrong way. Charge leakage from a floating gate or charge-trap layer rises with storage temperature, so a worn drive left in a hot car or an unconditioned storage room loses readable charge faster than the same drive kept cool.
The practical reading of JESD218 for a wear-failed drive is directional, not a countdown clock: retention is already shorter than the rated minimum, and every degree of extra heat shortens it further. That is the case for imaging a read-only drive sooner rather than storing it.
SSD Recovery Pricing
NAND wear recovery follows our standard SSD recovery pricing tiers. Most wear-related cases fall in the lower SSD tiers because the controller is typically functional and the data is readable through standard channels. Cases requiring PC-3000 firmware-level intervention or drives with controller lockout fall into the higher tiers. Free evaluation, firm quote before work begins. No data = no charge.
Low complexity
Simple Copy
Your drive works, you just need the data moved off it
Functional drive; data transfer to new media
Rush available: +$100
$200
3-5 business days
Low complexity
File System Recovery
Your drive isn't showing up, but it's not physically damaged
File system corruption. Visible to recovery software but not to OS
Starting price; final depends on complexity
From $250
2-4 weeks
Medium complexity
Circuit Board Repair
Your drive won't power on or has shorted components
PCB issues: failed voltage regulators, dead PMICs, shorted capacitors
May require a donor drive (additional cost)
$450–$600
3-6 weeks
Medium complexity
Most Common
Firmware Recovery
Your drive is detected but shows the wrong name, wrong size, or no data
Firmware corruption: ROM, modules, or system files corrupted
Price depends on extent of bad areas in NAND
$600–$900
3-6 weeks
High complexity
PCB / NAND Swap
Your drive's circuit board is severely damaged and requires NAND chip transplant to a donor PCB
NAND swap onto donor PCB. Precision microsoldering and BGA rework required
50% deposit required; donor drive cost additional
50% deposit required
$1,200–$1,500
4-8 weeks
Hardware Repair vs. Software Locks
Our "no data, no fee" policy applies to hardware recovery. We do not bill for unsuccessful physical repairs. If we replace a hard drive read/write head assembly or repair a liquid-damaged logic board to a bootable state, the hardware repair is complete and standard rates apply. If data remains inaccessible due to user-configured software locks, a forgotten passcode, or a remote wipe command, the physical repair is still billable. We cannot bypass user encryption or activation locks.
No data, no fee. Free evaluation and firm quote before any paid work. Full guarantee details. NAND swap requires a 50% deposit because donor parts are consumed in the attempt.
- Rush fee
- +$100 rush fee to move to the front of the queue
- Donor drives
- A donor drive is a matching SSD used for its circuit board. Typical donor cost: $40–$100 for common models, $150–$300 for discontinued or rare controllers.
- Target drive
- The destination drive we copy recovered data onto. You can supply your own or we provide one at cost plus a small markup. All prices are plus applicable tax.
SSD NAND Wear: Frequently Asked Questions
Can data be recovered from an SSD stuck in read-only mode?
What does SMART attribute 'Media Wearout Indicator' at 0% mean?
Why did my SSD suddenly become write-protected?
Does SSD wear affect all brands equally?
Can firmware repair fix a wear-related SSD failure?
Can I recover files I deleted from a worn SSD?
Is a worn SSD that shows 0 bytes failing in firmware or in the NAND?
How long does an SSD last before NAND wear becomes a problem?
Related services
Related SSD Recovery Pages
Full SSD recovery service overview
FTL and firmware module recovery
Controller panic and capacity loss
Firmware-driven premature wear
Interpreting SMART attribute warnings
Controller failure diagnosis
SSD in read-only mode?
Free evaluation. No data, no fee. Ship your drive to our Austin lab for professional imaging before further cell degradation.