Unplug your computer, and the gigabytes in your RAM vanish within milliseconds. Yet you can toss a USB thumb drive or SSD into a drawer for ten years with zero battery, plug it back in, and every single photo, document, and video is still perfectly preserved. How does a piece of silicon remember without electricity?
Flash memory stores data without power by trapping microscopic electrical charges (electrons) inside a physically isolated conductive pocket called a Floating Gate (or silicon-nitride Charge Trap). Because this pocket is completely encased in a solid glass dielectric insulator ($\text{SiO}_2$), the trapped electrons have no conductive path to escape. They remain locked in a potential well for over a decade, acting as a permanent electric shield that dictates whether the transistor reads as a 1 or a 0.
To appreciate why flash memory is such a marvel of modern engineering, you first have to understand why computer memory usually forgets.
Your computer's main memory (DRAM) stores bits inside microscopic capacitors. But a capacitor is like a leaky bucket — the electrons drain out within milliseconds. To prevent data loss, the memory controller must recharge every single capacitor hundreds of times per second (a process called memory refresh). Cut the power, and the charge leaks away forever.
Standard transistors (MOSFETs) have a single gate on top. When you apply positive voltage to that gate, it pulls electrons into the channel below, turning the switch ON.
In 1967, Bell Labs engineers Dawon Kahng and Simon Sze invented a game-changing modification: The Floating Gate Transistor (FGMOS). They inserted a second gate between the control gate and the silicon substrate, completely encasing it in silicon dioxide ($\text{SiO}_2$) — the exact same compound that makes window glass.
Glass is one of the greatest electrical insulators known to physics. Once electrons are pushed inside the floating gate, there are no wires, no contacts, and no circuits connecting it to the outside world. The electrons are stranded on a microscopic conductive island surrounded by an impenetrable wall of glass. Under normal conditions, it would take centuries for those electrons to leak out naturally.
This creates an obvious engineering paradox: If the insulating glass wall is so impenetrable that electrons cannot escape, how on earth do we get them in there in the first place?
The answer is Fowler-Nordheim Quantum Tunneling:
In classical physics, an electron without enough energy to climb over an insulator wall is permanently reflected. But in quantum mechanics, an electron behaves like a wave. When you blast the control gate with a massive +18V to +20V programming pulse, you create an electric field of over 10 million volts per centimeter.
This immense field severely bends the silicon dioxide energy bands, making the physical barrier ultra-thin. The electron wave passes straight through the forbidden solid glass barrier and traps itself on the other side. Every photo saved on your SSD literally relied on quantum tunneling!
How does the memory controller know if a cell holds a `1` or a `0` without letting the electrons escape? By measuring the Threshold Voltage ($V_{th}$):
One of the most important architectural rules of NAND Flash:
Why? To erase, you must apply a high positive voltage (\(+20\text{V}\)) to the P-well silicon substrate to pull electrons back out of the floating gates. Because hundreds of pages share the same physical silicon substrate well, you cannot erase a single byte or page in isolation without erasing the entire block! This is why SSD controllers need complex Garbage Collection and Flash Translation Layers (FTL).
In early SSDs, each transistor stored only 1 bit (SLC - Single Level Cell): empty gate = `1`, full gate = `0`. But today's multi-terabyte drives use TLC (3 bits) and QLC (4 bits per cell). How can a single transistor hold 4 different bits?
By precisely controlling the exact number of electrons trapped inside the floating gate!
Unlike a spinning hard drive platter that can be magnetized and demagnetized billions of times, every write to a flash drive causes irreversible physical damage to silicon.
Forcing electrons to quantum-tunnel through solid silicon dioxide under a 20-volt field is violent on an atomic scale. Over thousands of write/erase cycles, the high-energy electrons crack chemical bonds in the oxide lattice, leaving behind tiny atomic fractures called trap states. Eventually, the oxide layer degrades so much that electrons leak out through the fractures. The cell can no longer hold charge — it becomes a bad block.
To prevent your drive from failing prematurely, SSD controllers employ sophisticated algorithms:
Under normal room temperature conditions, a healthy flash drive or SSD will safely retain data for 5 to 10 years without power. However, high temperatures dramatically accelerate electron leakage (thermally assisted tunneling). An SSD stored at 55°C (130°F) in an attic can lose data in less than a year!
In modern high-capacity SSDs, planar floating gates ran into a physical limit: cells were so close together that electrons in one gate were capacitive-coupling into adjacent cells! Manufacturers switched to Charge Trap Flash (CTF), which replaces the conductive polysilicon floating gate with an insulating silicon-nitride layer, and stacked them vertically up to 200+ layers tall (3D NAND).
No! Traditional hard drives store data using magnetic domains, which can be erased by a powerful rare-earth magnet. Flash drives, SSDs, and SD cards store data using electric charge (trapped electrons), which is completely immune to normal magnetic fields.