There are plenty of things that you can’t explain with just electron flow: You have to consider fields, too. But then, if one of the components in your circuit is a hamster wheel powering a generator, then electrons and fields aren’t enough, either, and you also have to consider things like metabolism and muscular action. So what? That doesn’t make electron flow in any sense wrong.
The physics of what happens to the mobile charges in an electric circuit can get complicated.
Consider a simple flashlight. You press the switch and the light turns on. Electrons are flowing in the wires. Simple enough, right? In the LED, however, electrons flow and holes flow. What happens to the holes when they get to the copper leads of the LED? Do they disappear? And what creates them in the first place? And things get really weird at the battery, because the electrons on the wires do not enter or exit the battery; there are no mobile electrons in the battery.
If you want to know more about this stuff, get some books on solid state physics and electrochemistry.
Most EEs don’t need to know this low-level stuff in order to design things. Much like most MechEs don’t need to understand quantum mechanics in order to do their work. Us EEs talk in terms of voltage, current, impedance, power, etc. We don’t care much about electrons.
Seriously. Explain how semiconductors or vacuum tubes work. Electrons flow.
Where we have analogies like water flow or BB flow we are only modelling the aspects of the physics pertaining to flow of charge, and short range interactions. But it is enough to get a long way there. So much so that low frequency behaviour of electrical circuits can be modelled quite successfully. This includes transmission lines and resonant behaviour. But since these models don’t model the external fields of a wire they don’t model antennae and EM radiation.
Perhaps the critical point to note is that in these models, neither water nor a chain of ball bearings is incompressible. An impulse travels through either of these media at the speed of sound in that media - not at the gross speed of movement of the media in response to a force. Take a pipe full off ball bearings. Push on one end, the BB on the other end responds after the impulse has travelled, and the seed of sound in the BBs. Same with electrons. Where the simple analogue breaks down is that we didn’t add any restoring force to the chain of BBs. In a wire a displaced electron is pulled back into the wire due to the lattice filled with positively charged nuclei. So it won’t pop off the end of a wire until it gets enough potential to overcome the surface attraction.
Electrons can of course actually pop off. This is how vacuum tubes or other thermionic devices work. Or those negative ion generators that were in vogue a few decades ago. Heat a conductor up enough and the free electrons start to move faster and faster as the rest of the lattice also vibrates faster. As in any material there is a distribution of velocities for the electrons, and a proportion of the electrons will eventually have velocities fast enough and in a suitable direction to break free of the surface. Put it in a vacuum and place a positive charge nearby and you can stream electrons across the gap. This is what occurs in any vacuum tube. The physics of what is occurring and how you can derive the electrical properties of something like a real work triode are really neat. The characteristic amplification curve of a triode is directly related to the geometry of how thermal electrons are emitted. The energy needed to break free varies by material, and is characterised by its work function. You can also liberate an electron from the surface of a conductor with an incident photon of sufficient energy to overcome the work function. (JJ Thomson got a Nobel for thermionic emission, and Einstein got one for the photoelectric effect.)
However emission of electrons has nothing to do with radiation of EM waves. Electrons don’t exit an antenna. EM waves do.
Electrons can bunch up anywhere. The trivial example can be found in common electrostatic experiments. The easiest of which is to just create a static charge.
An example of a static charge and conductors is a Van Der Graff generator. Electrons are pumped into the metal sphere by the running belt. The metal sphere carries an excess of electrons, and as more electrons are deposited into the sphere the charge grows, and the electric potential between the sphere and the source of electrons grows. Eventually the potential exceeds the breakdown voltage of air and the air ionizes and creates a conductive path that discharges the sphere, and so on. Electrons travel along the discharge.
Electrons do bunch up. Why would they not? There are free electrons in metals, that is what makes them a metal. If electrons are bunched up in one part of a metal conductor, there is a net electrostatic potential, from that part, relative to parts of the conductor where the electrons are less bunched up.
However once we get to transmission lines we need to include not just the electrostatic force tied to the location of charge carriers, but must also start to worry about the magnetic field tied to moving charges. As the electrons are disturbed, they move, and any moving charge results in a magnetic field. So, apply a potential to the ends of a wire, electrons start to move in response, a magnetic field starts to arise due to the moving charge encircling the conductor. This actually takes some energy to do. If the potential is removed from the wire, the electrons will relax back to their equilibrium locations, resulting in a current in the other direction. This results in a reversing magnetic field, something the existing magnetic field opposes, and as the existing field collapses it induces a current that opposes the change in potential. When you have a wire pair, all of this can be usefully captured as the infinite set of capacitance and inductors mentioned earlier. The wires are inductive (they create a magnetic field if a current is run down them). A pair of wires is mutually coupled by both an extended inductive coupling, and an extended capacitive coupling. In addition the wire itself is an extended resistance.
We end up with a mutual interplay between potential field and magnetic field. A signal travelling down a transmission line has an electrostatic wave and a magnetic field wave. Critically, these two are out of phase by 90 degrees. If you sit watching a bit of the line - as the current in the wire, and the associated magnetic field reach a maximum - the potential field at that point is reaching zero, and vice versa.
The transmission line presents a particular characteristic impedance for signals that have a wavelength that is of a similar order to the length of the line. TL effects can be seen at say 1/20 of a wavelength if you look hard, and are well established by about 1/10. Any wavelength longer than this and the line tends to look like a simple pair of wires. If the TL is terminated with an impedance other than the line’s characteristic impedance, things get out of whack. In the extreme this could be a short circuit or an open circuit. Cutting to the chase, their can be reflected energy, and the wave travels back down the line. A superposition of the waves is not an issue. The phase of the magnetic and electrical fields remains, and that allows superposition.
If you just look at the model of capacitors and inductors you get the same behaviour.
So electrons do flow then?
However, be careful here. Quantum tunnelling is working in terms of the wave function of the electron. This wave has nothing to do with the EM wave that the electron may be part of. You can state tunnelling in terms of Heisenberg uncertainty and the wave function. Tunnelling also has nothing to do with the base question or EM reflections and improperly terminated coax. Also, that electron wave function only ever describes an electron, there is no “better particle” that you can turn the wave back into. We can work in terms of pseudo-particles, and these also have wave functions. But you don’t get to turn an electron into a different particle. You can turn an ensemble of real particles into an ensemble of the absence of them. But that is more a calculation convenience. Electrons are real particles, holes are a convenience.
QM does cover normal flow of current in a conductor, and when it does, it can introduce another pseudo-particle. As has been described, electron flow carrying a signal looks a lot like an acoustic wave propagating in a medium, with the density of particles changing as the wave propagates. Since we have waves in the conductor, we can note that QM allows these waves to be analysed in terms of their pseudo-particle QM counterpart. This is the phonon. In the same manner as a photon mediates force between electrons, in a conductive medium, phonons can be used. The difference being that photons are real, while phonons are not. It is annoying that there is only a one letter difference in the name. But it is a good name. Phonons can be usefully used to model Cooper pairs and superconductivity.
I respect your knowledge of the field.
We are discussing the suitability of various analogies that are used for the purpose of illustration. Vacuum tubes can be explained using fields as well as particles or electrons. But then you get both, like when particles actually came off of the cathode and caused problems.
I always had difficulty with the electron flow rationalization of Williams tube digital storage. The idea that slurring more electrons in to the pit would write a zero where fewer would write a one did not make sense. It seemed more reasonable that thinking of it as slurring a field would dissipate the charge better than adding electrons. And, that’s the problem. If I get focused on the ‘reality’ of the analogy I’ll get stuck and miss the point. The point was that electron beam storage worked and the control circuitry broadened the X position pulse to write a zero.
I can explain the high inductance of a two element neon bulb with electron flow, but not it’s negative resistance property. In the negative resistance region it will amplify just like a transistor but with an entirely different rationalization and design approach. I’ve never tried to defend an electron analogy for negative resistance. The equations are sufficient.
One bit I found interesting working with radio was that you can switch between signal transmission and field models (somebody mentioned VSWR above). That you could model a signal either as traveling (forward and reflected) waves, or as a varying total voltage/current. They are mathematically equivalent, but you switch models & analogies based on which perspective is most useful for your task.
Never got the hang of Smith charts, but I loved my network analyzers.
I assume you are referring to the avalanche breakdown region of a gas discharge lamp’s operation. I would have thought that electron flow was probably the best way of describing how the effect occurs.
Overall I think that common tactics in teaching the physics and engineering aspects of electrical systems need to be very clear about what models are in use. It can become confusing for students when the models seem to jump from one to another without apparent reason, and leave one with lingering doubts and often bad mental models. Perhaps the most common being how one aligns a water flow model with an open circuit. Why doesn’t the current just run out onto the floor like a cut water pipe would?
I think my high school physics teacher did me a favour very early on when he pointed out how few electrons are needed to obtain a macroscopic electrostatic effect. Then one realises just how little distance electrons actually need to move in a conductor for useful effects to become apparent. Then it makes a great deal more sense. One electron volt is actually a substantial amount of energy.
Engineers get taught the most useful abstraction for any given task, with any unnecessary complication removed. Which is great when your job is to apply known physics in a known situation. Most circuit analysis can be done with little more than Thévenin or Norton equivalent boxes L,C,R and Ebers-Moll transistors and ignores the existence of magnetic or electrostatic forces or the existence of electrons. Digital designers get away with vastly more.
OTOH, if one wants to understand, and not just apply, the question is much harder. Nobody is expected to apply QED to designing a radio antenna. But if you want to understand the underpinning physics at play, going down a few layers helps greatly. As a pedagogical ideal I have always tried to present any model in two contrasting ways, and suggest which is the more convenient for our purposes. A bit of compare and contrast, and indicate which is closer to the truth. So much of applied physics is what has become generally know as an emergent behaviour. Understanding that we are applying a simplifying model to an emergent behaviour, IMHO, makes it clearer what we are doing. Minds seem to learn better with a bit more of he landscape, that than a simple dry model that seems disconnected with reality in weird ways that are never explained. The edge cases are where gaps in understanding become apparent.
I worry how many students walk out of school still wondering why electrons don’t pour out of the end of a wire.
Given the Veritasium video mentioned above it becomes even more worrying. Derek mentions in the video that he taught students a totally weird model electrical flow, one that was wrong. He then proceeded to tell us what the real model was, and got that badly wrong as well. Yet he has a PhD from the physics department of a major university, focussing on the topic of physics education and misunderstandings of physics at the undergraduate level. There is a weird irony here.
Thanks for the informative response.
Once, at IBM, I submitted a patent disclosure that involved an array of particles suspended in a vacuum chamber. The Physics Phd on the review board looked at it and said “If there are particles in it then it’s not a vacuum” and I said “There’s an electron cloud in a vacuum tube”. He grunted something about analogies as he stamped REJECTED on my brilliant proposal.
… blink. Isn’t that the easy question? Cut pipe is a short circuit - pump as much “water” as you want. An open circuit being a sealed pipe - infinite resistance to flow.
I have had discussions with a mechanical engineer that were annoying because of our unexamined assumptions about what “open” means. As an EE, I think of open as implying that no flow is possible, but of course the MechE had a different idea. I’d analogize an open electrical circuit as a impenetrable, inflexible blockage in a water flow circuit, but I suppose you could also analogize it as a cut pipe, which also rapidly leads to no water flow in the pipe because there’s no water in it anymore.