BrightKidz Library
Subjects
Three flash memory cells, two of them holding trapped charge Three identical memory cells stood side by side like exhibits in a case. Each one is a thick block with a small sealed room hollowed out in the middle of it, a bar above and a bar below, none of them touching. The left cell and the right cell each have three small pellets shut inside the sealed room. The middle cell's room is empty. Nothing joins the rooms to the outside, which is why what is inside them stays there.

Why Flash Memory Remembers With the Power Off

About 11 minutes

You already know the bargain behind the three kinds of storage, and you know how a hard drive holds on to a file: magnetism, on a spinning disc, sitting there whether the power is on or not.

Now look again at what was said about the other two. Information in an SSD or a USB stick is "held electronically". That is not really an answer. Electricity is the thing that stops when you pull the plug — so how can a file made of electricity survive being switched off?

Electricity does not need to be moving

The word "electricity" usually makes people think of something flowing: a current running down a wire, going somewhere.

But an electric charge does not have to go anywhere. Rub a balloon on your hair and it sticks to a wall. Nothing is flowing. There is simply some charge on the balloon that has nowhere to go, and it stays put until it finds a way out.

That is the whole idea behind flash memory. It does not keep a current running. It keeps a small amount of charge stuck.

A room with no doors

Inside a flash chip there are billions of tiny cells. Each one is built in layers, and the important layer is a small patch of metal called the floating gate.

It is called floating because it is not joined to anything. It is walled in on every side by a material that electricity cannot cross — the same reason the plastic round a cable keeps the current inside. There is no wire in, no wire out, no path anywhere.

So if you can get some charge in there, it has no way of leaving. Not when the power goes off. Not when the computer is thrown in a drawer for a year.

So how does anything get in?

Here is the part that sounds like cheating, and is not.

Given a big enough push — a much higher voltage than the chip uses for ordinary work — some charge can be forced straight through the wall and into the floating gate. It is not going round the wall or through a gap. It is being shoved through a barrier that would normally stop it.

Once it is through, the push stops, and the wall goes back to being a wall.

Step through one cell being filled, left alone, and read.

Writing, keeping and reading one flash memory cell One memory cell drawn in cross section, shown in four stages. The cell has a control gate bar across the top, a thick wall block below it with a sealed floating gate room hollowed out inside, and a wire along the bottom. First the room is empty. Then a strong push drives four pellets of charge in through the wall and into the room. Then the power is switched off and the pellets are still sitting there, because the wall has no way out. Finally a current runs along the bottom wire, and what is trapped in the room changes how easily that current gets through. control gate the wall floating gate wire 1. Nothing inside. This cell is a 0. a big push 2. Charge is forced in through the wall. power off 3. Power off. The wall keeps them in. 4. The trapped charge changes the current.

Four stages in the life of a single cell. Watch stage 3 especially — it is the one where nothing happens, and that is the whole point of it.

  1. The floating gate is empty. Read this cell and you get a 0.
  2. A high voltage on the control gate forces charge through the wall and into the floating gate.
  3. The power is switched off. The charge is still in there, because there is nowhere for it to go.
  4. To read the cell, a current is sent along the wire underneath. Charge sitting overhead makes that current harder to push through — and how hard it is to push through is the answer.

Stage 4 is worth reading twice. The computer never looks inside the floating gate. It cannot. It asks a question about the wire underneath, and the answer to that question depends on what is trapped above.

A 1 and a 0 are not two different things stored in the cell. They are the same cell, easy or hard to push a current through.

Why this explains the price

Both facts you already know about SSDs come out of this.

Why it is fast. There is no arm and no disc, so there is no waiting for anything to spin round. Reading a cell means putting a voltage on a wire and seeing what happens, and every cell is the same distance away — which is to say, no distance at all.

Why it costs more. A hard drive stores more by adding another metal disc. A flash chip stores more by having more cells, and every single cell has to be built with its own walled room, at a size far too small to see. The disc is cheap because it is one plain object. The chip is expensive because it is billions of tiny ones.

The catch nobody mentions

Forcing charge through a wall damages the wall a little. Not much — but a little, every single time.

After enough writes the wall stops being a perfect wall, and the cell can no longer hold its charge reliably. Flash memory does not last forever. It lasts a certain number of writes.

This is why an SSD quietly spreads your writing around instead of using the same cells over and over, a habit called wear levelling. If your computer kept rewriting one cell, that cell would wear out while billions of others sat unused. So it does not. It keeps moving on.

A USB stick sits in a drawer for a whole year with no power at all. Why is the file still on it?

Next time you pull a USB stick out without thinking about it, that is what you just did: shut a few billion tiny rooms and walked off with them in your pocket, each one still holding the pinch of electricity somebody forced into it.