Why Your Solid State Drive Failure Is Going to Be Horrible

5/5 - (1 vote)

Your computer’s inevitable solid state drive failure is going to give you a very bad day. Maybe. To create amazingly fast solid state drives, engineers had to exploit quantum mechanics. But, it comes with a cost.

Quantum Mechanics and Hard Drives

Quantum mechanics is all around us. Most of the time, we are not aware of its effects.

Solid state drives (SSDs) are similar. They look like ordinary drives. They work like ordinary hard drives. But they exploit quantum mechanics to store more data and access it faster. (And, thankfully, they don’t sound like old-school harddrives.)

Engineers make SSDs from two different kinds of storage components:

Floating gates, and
Charge traps

These components work similarly. The storage component is fully isolated from any conductor. The storage itself is physically isolated from the gate and channel. Electrical current can’t seemingly affect it.

So, how do you trap the charge inside the storage part?

Here comes quantum mechanics and quantum tunneling. Sometimes, this is better known as the Fowler-Nordheim tunneling. Well, to be precise tunneling erases SSD memory. And a separate process, known as hot carrier injection, writes it.

What is Tunneling?

In quantum mechanics everything is about probability.

In your everyday life, if you throw a ball at a wall, the ball bounces right back.

But a quantum ball has a small chance of passing right through the wall. Quantum mechanics dictates there is a small probability of finding a a ball inside solid object. The chance of a ball moving past a wall does not drop to zero at the edge of a wall. But, it does drop exponentially quickly. So if object is small enough, there is a chance of of a small particle passing through it!

And this happens in a SSD storage element. Electrons try to pass through the insulator right into the storage part. Though, no circuit connects the two components. Enough electrons pass through it. But the flow is weakened.

Hot Carrier Injection

Hot carrier injection stores data in a storage element. Basically, you crank up a high voltage source and give enough kinetic energy to electrons. Those electrons blast through the insulator into the storage component.

Every MOSFET transistor has 3 pins, Gate, Drain and Source. Gate is control and controls current flow between Source and Drain. To program FLASH bit high voltage is applied on Gate, lower voltage on Drain and Source and current is let to flow between Drain and Source. High voltage on Gate gives flowing electrons enough energy to inject into Storage Element (floating gate or charge trap). Once there, electrons stay trapped. Before new data is written old one must be erased. Erasing Storage Element is removing trapped charges. This is done using tunneling. Again, high voltage is applied on gate while drain and source are grounded. High-voltage gives the trapped electrons enough energy to tunnel backwards through the insulator.

The size of these components is extremely small, about 10 nanometers. That’s enough to keep electrons trapped inside the storage component. But the size is thin enough to cram them into the component at high voltage. They balance just right.

Solid State Drives have a fatal flaw

But there are problems. High voltages used for erasure and programming affect these nanometer-sized structures and weaken them over time. Overtime, thin insulators loses its resistance and charges can no longer be trapped. Charge starts leaking out. They no longer work.

Unfortunately, modern QLC FLASH can only be rewritten about 200 times. For example, Corsair MP600 1TB has only 200 TBW. TBW, or Terabytes Written, is number telling us how many times SSD could be erased and rewritten.

200 TBW sounds like a big number. But EVERY write requires previously written cells to be erased. As an SSD fills up, the amount of free space drops. A shortage of free space causes extra wear and tear on certain commonly used blocks. Today, SSD controllers contain different “wear leveling” algorithms to try to avoid this. But no matter what, free blocks reach their erasure and rewriting limits. It’s inevitable.

SSDs are convenient, but they are very bad storage medium and can lose data quickly. Because of the small sizes and data corruption problems, very complex data correction algorithms try to protect data. Even brand new SSD loses data and software tries carefully to recover. These modern drives use up to 20% of their space just for this error correction.