Flash memory stores each bit by trapping electrons on an electrically isolated "floating gate" inside a transistor. Fully insulated, that charge has nowhere to leak, so no power is needed to hold it in place. The trapped charge shifts the transistor's switching voltage, which the controller reads as a 0 or a 1.
The Floating-Gate Transistor: A Cage for Electrons
A flash memory cell starts as an ordinary transistor, then adds a second, buried gate sandwiched between two layers of insulating oxide, typically silicon dioxide. That buried layer is the "floating" gate — floating because it has no wire connected to it at all. Whatever charge ends up trapped there stays there, blocked in every direction by insulation, until something deliberately forces electrons across the barrier. This is the fundamental difference from a DRAM capacitor, whose charge leaks away in milliseconds and must be constantly refreshed: a floating gate can hold its charge for years with zero power applied, which is what makes flash a genuinely non-volatile technology.
Writing: Forcing Electrons Onto the Gate
Getting electrons onto an insulated island sounds difficult, and it is — it takes a relatively high voltage to do it. Flash cells use one of two quantum-mechanical tricks to push electrons through the insulating barrier: Fowler-Nordheim tunneling, which applies a strong electric field across the oxide until electrons tunnel through it, or hot-electron injection, which accelerates electrons hard enough to leap the barrier directly. Either way, once the write voltage is removed, the electrons are stuck on the far side of the insulator with no way back — that's the stored bit.
Erasing: Pulling the Charge Back Off
Because writing floating-gate cells requires a comparatively strong voltage and stresses the thin oxide layer a little each time, flash memory doesn't erase a single bit at a time. Instead, chips erase in large groups called blocks, applying a reverse voltage that pulls the trapped electrons back off every cell in the block simultaneously. This block-erase requirement is the root cause of a whole category of SSD behavior — why drives need to be told which data is no longer needed so they can group live data separately from stale data before erasing, a housekeeping process covered in our guide to TRIM. Every erase cycle wears the oxide slightly, which is also the physical root of flash memory's finite lifespan.
Reading: Sensing a Shifted Threshold Voltage
To read a cell, the controller applies a specific voltage to the control gate and checks whether the transistor switches on. Electrons sitting on the floating gate act like a shield, raising the voltage the control gate needs to overcome before the transistor conducts. A cell with electrons trapped on it needs more voltage to switch on than an empty one — the controller senses that difference and reports a 0 or a 1 accordingly. Modern drives store more than one bit per cell by parking the charge at one of several intermediate levels rather than just "empty" or "full," a technique covered in our companion piece on SLC vs. MLC vs. TLC vs. QLC NAND.
Where You Meet Floating-Gate Flash Every Day
Every device that keeps data with the cord unplugged relies on some form of this cell. A compact drive like the SanDisk Ultra Flair 128 GB USB flash drive is floating-gate NAND in its simplest packaging — a controller and a handful of flash dies in a keychain-sized shell. A SanDisk Ultra 256 GB microSD card uses the identical cell technology shrunk into a card the size of a fingernail for cameras and drones. And the internal drive running your operating system, such as a Crucial BX500 480 GB SATA SSD, is built from the same floating-gate cells, organized into far more pages and blocks and paired with a much more sophisticated controller managing wear, error correction, and the TRIM housekeeping mentioned above.
Frequently Asked Questions
Why doesn't flash memory need power to keep data?
Because the electrons that represent a stored bit sit on a gate that is fully surrounded by insulating oxide. With no conductive path in or out, the charge has nowhere to leak to, so it stays put indefinitely without any power applied.
Does flash memory wear out?
Yes. Each write and erase cycle pushes electrons through the insulating oxide layer, and that stresses it a little each time. Enough cycles eventually degrade the oxide's ability to hold charge cleanly, which is why flash drives carry finite write-endurance ratings.
Is flash memory the same thing as RAM?
No. Both store bits as electrical charge, but RAM's capacitor-based cells lose their charge in milliseconds without constant refresh, while flash's floating-gate cells hold charge with zero power for years. That's the core tradeoff between the two families of chips.
Why can't flash memory be erased one bit at a time?
The erase voltage that pulls electrons back off the floating gate is applied to an entire block of cells at once by design, both to simplify the chip's circuitry and because erasing is the more stressful operation — doing it in bulk, less often, extends the memory's usable life.
Share this article: Twitter