How many Electrons make a bit?

Question

A bit, correct me if i'm wrong, is a digital signal which can be 0 or 1. To determine its value, a processor awaits an X amount of time. My question is: is there any way to determine how many electrons passed through the processor to make that bit?
If a Z amount of electrons make a 0, a single electron, added to that Z amount, would change the bit's value from 0 to 1 (correct me if i'm wrong).
So, can we exactly determine by how many electrons is Z made of?

"To determine its value, a processor awaits an X amount of time." - not really, the voltage at a point is compared against a threshold, at regular intervals. Thinking in terms of "number of electrons" isn't particularly useful. — pjc50, May 27 '19 at 13:18
A bit is a unit of information, not a quantity of electrons. A bit can be a 1 or a 0 written on a piece of paper if you want it to be! — Hearth, May 27 '19 at 13:22
Many FETs gates are 10femtoFarads or less. At 1 volt, you need 1e-14 Farad * 1.6 e+19 electrons/couomb, or 160,000 electrons. — analogsystemsrf, May 28 '19 at 00:18

score 4 · Answer 1 · answered May 27 '19 at 08:59

So, can we exactly determine by how many electrons is Z made of?

Simple answer: No.

More elaborate answer:

A "bit" is just information. In electronics we usually deal with currents and voltages and not electrons. This is much simpler than dealing with electrons. So we don't use a single or a certain amount of electrons to represent a bit.

Usually in a CPU a bit is represented as a voltage, a low voltage means "0" and a high voltage means "1". How many electrons does that involve? No one knows. Also: it is irrelevant unless you would use really tiny structures which can work with a single electron.

And even if you would find out how many electrons are involved in a certain operation in a certain location in a certain CPU, at the moment anything is changed like:

temperature
CPU (so same CPU model but a different one)
Different location in the CPU
different supply voltage of the CPU

the number of electrons involved will be different.

And in the end, that doesn't matter as we use currents and voltages which are an effect caused by electrons but involving many of them. Even a small current like 1 nA means that around \$6.2*10^{9}\$ electrons are passing each second.

And if the signalling rate were 6.2 * 10E9 bps, does this imply one electron per bit? How about if the signalling rate were higher, would you have a fractional electron? — Phil Freedenberg, May 27 '19 at 16:10

score 0 · Answer 2 · answered May 27 '19 at 08:56

To give a absolutely precise answer would be difficult because of so many variables to consider. These variables would be cross-sectional area of the conductor inside the microprocessor , amount of current that is being switched on and off as well as the duration of the pulse. The only one you would have to calculate is the 1 because it is the ON state and the current is flowing. The amount of electron flowing in 0 state would be 0 electrons.

The exact amount if all the parameters were known would probably involve a lot of calculus and some expertise from the physics site on stack exchange.

score 0 · Answer 3 · answered May 27 '19 at 09:19

Electrons are also not the only charge carriers and they don't even move all that fast. The charge moves through your circuit faster than an electron can.

I know it's defined as a 'flow of electrons', but individual free electrons aren't really a meaningful unit.

score 0 · Answer 4 · answered May 27 '19 at 10:54

The data sgeet will tell you voltage and bias currents needed to reliably signal the processor.

Sure it's possible to push into the zone where the datasheet says the value is indeterminate and still get consisten results, but when your signal is in that range the result could be effected by nearby signals inside the chip, magnetic fields, the temperature, supply fluctuations, manufacturing differences, etc.

so yes, \$Z\$ can be measured experimentally but don't expect a fixed number for \$Z\$, it could be different in 10 seconds, or tomorrow, or with new batteries and \$Z\$ probably fluctuates with each test...

score 0 · Answer 5 · answered May 27 '19 at 12:44

To add to the answer of @Bimpelrekkie

The active(high) and inactive (low) states of any CPU or microcontroller, not only they vary with the power supply or CPU but there is also a wide range where some voltage is considered high or low. For example, if a CPU runs on 1.2 Volts. 0.0 volts to 0.3 might be considered low. And 0.7 to 1.2 Volts considered high. Which would considerably change the amount of electrons. So even for the same CPU and same Power supply etc.. Every internal connection in a CPU might have a slightly different electron flow.
Which adds to what already was impossible another impossible.

score 0 · Answer 6 · answered May 27 '19 at 13:55

can we exactly determine by how many electrons is Z made of?

We don't count electrons in electronics, rather we count charge Q=CV.
All FETs have an input and output charge associated with switching states. On discrete parts, specs are included for Q, Ciss,Coss and others.

If we consider changing the state of a single 5V CMOS input rated at 2pF input, we can estimate $$Q = CV = 10 ~pC.$$

The energy to change states here can be estimated as E=1/2CV^2= 25 pJ.
The peak noise current depends on the driver impedance where RdsOn ~ Vol/Iol.
The transition voltage does have a certain amount of shoot-thru current in the CMOS driver but the rise time is limited by the driver impedance and current limit where at some point T=RdsOnCin such that Ipk=CdV/dt is approximately Ipk=~Vdd/RdsOn for some time T.
- the approximation depends on logic voltage range and Vt threshold where C is nonlinear.

score 0 · Answer 7 · answered May 27 '19 at 14:31

0

If it is a spintronic processor the answer can be "none"

answered May 27 '19 at 14:31

Dirk Bruere

13,723
9
54
115

How many Electrons make a bit?

7 Answers7