Updated on
NAND flash is the nonvolatile memory used to store data in SSDs, USB drives and many other devices. Its cells retain information without continuous power. Terms such as TLC, QLC and 3D NAND describe different parts of its design.
How NAND flash stores a bit
A flash cell stores electrical charge in an insulated storage structure. Changing that charge changes the transistor’s threshold voltage: the voltage needed for it to conduct. During a read, the circuitry compares the cell with reference levels and interprets the result as data.
A one-bit cell needs two distinguishable states. A two-bit cell needs four, a three-bit cell eight, and a four-bit cell sixteen. The name NAND refers to the arrangement of the cell circuitry. Describing a stored bit as simply an open or closed logic gate misses the charge and threshold-voltage mechanism. KIOXIA’s multi-level cell explanation illustrates the relationship.
SLC, MLC, TLC and QLC compared
| Type | Bits per cell | Required states | Main trade-off |
|---|---|---|---|
| SLC | 1 | 2 | Wide margins between states, with relatively low storage density. |
| MLC, in common two-bit usage | 2 | 4 | More data per cell than SLC, with narrower margins. |
| TLC | 3 | 8 | Higher density, widely used in consumer SSD designs. |
| QLC | 4 | 16 | Still higher density, with tighter programming and sensing requirements. |
More bits per cell can lower the cost of storing a given amount of data, but the smaller margins make endurance and programming behavior more demanding. A newer QLC product does not inherently wear less than TLC. Comparing complete drives requires their specifications and the workload, rather than the cell acronym alone. Kingston’s NAND comparison explains these broad trade-offs.
How 3D NAND differs from TLC or QLC
Planar NAND places cells in a horizontal arrangement. 3D NAND builds cells in vertical layers, increasing the number that can fit in a given footprint. Layer count describes this physical organization. Bits per cell describes how much information each cell stores.
A drive can therefore use 3D TLC or 3D QLC. More layers do not automatically mean a faster or more durable SSD. Manufacturing generations, cell design and controller choices also matter. There is no useful universal layer-count ceiling to apply to every current drive.
Why NAND reads and writes pages but erases blocks
A NAND device groups cells into pages, then groups pages into larger blocks. Reading and programming happen at page granularity, while erasing happens at block granularity. The exact sizes depend on the flash design; there is no single page size for all SSDs.
This is why changing a file is more involved than overwriting the same flash location repeatedly. In a managed SSD, the controller typically:
- Writes the new data to an available page.
- Updates its mapping so the computer’s logical address points to the new location.
- Marks the old data as no longer needed.
- Later moves still-valid pages out of a block and erases that block so its space can be reused.
The cleanup is called garbage collection. It can create extra internal writes, which helps explain why sustained writing and available free space affect SSD behavior. Micron’s SSD technical glossary covers these flash-management tasks.
TRIM lets a supported operating system tell the drive which logical data is no longer needed. It helps the controller manage free space; it is not a promise of immediate physical erasure or a substitute for a secure-erasure procedure.
Why the SSD controller matters
The controller translates the computer’s logical addresses into locations in flash. It also manages error correction, wear leveling and the movement of data as flash is reused. These tasks help make the underlying cells usable as a storage device.
Some drives use a portion of their NAND in a one-bit mode as an SLC cache. Short transfers can be fast while that cache has space, then behave differently during a long write. This does not change all the drive’s NAND into permanent SLC. Kingston’s SSD and NAND guide covers the controller and endurance context.
Mapping information may use onboard DRAM or, on supported NVMe designs, Host Memory Buffer. HMB is a use of system RAM, separate from the flash that holds your files.
What to check when choosing or using an SSD
- Compare the exact model and capacity. Different capacities within a family can have different behavior.
- Check rated write endurance, such as TBW, together with warranty conditions. A rating is not a predicted date of failure.
- Look for sustained-write results relevant to your workload, especially for frequent large transfers.
- Keep important files backed up. Nonvolatile storage does not guarantee indefinite retention or prevent device failure.
For an SSD you are preparing for Windows, the separate MBR versus GPT choice concerns its partition layout and boot compatibility. It does not determine the NAND type or make the flash itself faster.
Frequently asked questions (FAQ)
Is NAND the same as RAM?
No. NAND keeps stored data without continuous power, while ordinary working DRAM needs power and refresh. That makes NAND useful for files and DRAM useful for fast working memory.
Does 3D NAND mean three bits per cell?
No. The 3D label describes stacked cell layers. TLC means three bits per cell. A product can use both descriptions at once.
Does NAND keep data forever without power?
No. Nonvolatile means the cells do not need continuous power to retain data. It does not guarantee unlimited retention or prevent a drive from failing. Keep important files backed up.