transitor: the base pin.

[free_electron]

Here is a reference that goes over the theory in some detail: http://ecee.colorado.edu/~bart/book/book/chapter5/ch5_1.htm

excellent reference. he really throws in the kitchen sink ( including things like 'early effects' and all the other oddities most people have never heard about)

now to come back to the debate.

for the sake of clarity can we define a few things ?

- an electron has a charge
- the core (nucleus) of an atom also has  a charge
- for a condustor ( and most atoms): the charge of the electrons cancels out the charge of the nucleus so the net charge is zero.
- a moving charge is a current .
- the number of charge carriers per second is a quantity we call ampere.

right ?

now, before we delve into transistors let's take a look at something more simple.

: an electric conductor , simplified, a wire. we will start with an ideal wire first ( no resistance)

A wire is made from a material with movable electrons. a wire' in rest ' has enough electrons  to fill the outer shell of all the atoms in the wire.
a wire in rest has no charge across it. There are charges moving internally as the electrons spin around their nuclei but, since the sum of the electron charge is equal to the sum of the nuclei charges the net effect is zero. it all cancels each other out.

now, let's push in an electron on side of the wire shall we ?
what happens : we unbalance the system. there is now an extra, unbound electron, in this stub of wire. so something has to give... none of the existing atoms want anything to do with this electron. so , as soon as there is a pathway for it to be shoved out of the wire it will be.
close the loop and the wire will spit out an electron for every electron shoved in the other end.

as electrons are charge carriers and moving charge carriers is called 'current '.... you got the idea right ?
of course in order to get an electron to move we need to apply a field.

now. step two. let's take a resistor. ( or , a not so ideal wire if you would like )

the principle still stands : one electron in is one electron out. electrons are particles , matter. laws of physics state you cannot create or destroy matter. ( unless in nuclear reactions). if i send 20 electrons in that resistor , 20 will come out. none extra are created and none are destroyed in this process.

Electrons are accelerated by the applied electric field. strenghten the field ( up the voltage) and you up the acceleration. Electrons only move at the speed of light in a vacuum. in material the run slower.

so an electron entering the resistor is accelerated by the applied field. so it gains kinetic energy. as it starts traveling into the wire it will sooner or later hit an atom and lose some of its kinetic energy. enrgy also cannot be created or destroyed , only converted. The electron is still propelled by the initial push it got from the external field so it keeps going towards the exit , all the while slamming into several atoms and losing some of its kinetic energy to the atom. this causes the atoms of the wire to jiggle faster thus producing ... heat !

but, eventually all twenty electrons sent in , will exit as none of the atoms int he resistor want them ! their charges are in balance.

In short we have been dealing with materials called 'conductors' : materials where the charges are in balance. Every proton has an electron so the net charge  is zero. shoe in an electron and it needs to come out.

Now it gets complex :Semiconductor.

a semiconductor is a material where the charges, for a single atom, are NOT in balance. there are not enough electrons to satisfy the number of protons (sometimes called holes) in the core.

Now, atoms are pretty clever little buggers and, in an INTRINSIC semiconductor, they will arrange themselves in a crystal like lattice so they can 'share' some of the electrons on their outer shell. As the electrons are spinning they no longer run around a single core but actually run in figure eight patterns around two cores. so now the atom cores 'see' enough electrons.

if you take a such a material and shove in an electron on side , one will roll out the other side. in an intrinsic semiconductor material the charge is not balanced absolutely ( if you were to count the number of electrons and number of protons there is a mismatch ) , but on a per-atom level they are balanced. as each atom has enough electrons for its number of protons. just not all the time  as the electrons keep circling mulitple cores.   note : this is heavily simplified. in reality it is more a brownian motion , 'electron soup' if you will.

a lump of intrinsic semiconductor material has no net charge and behaves like a wire. shove an electron in and one will come out. there is no need for the extra electron. the atom pairs sharing have no use for it. Sure, one of the two atoms sharing a single electron could absorb it and be content but then the other atoms is left with an electron missing so it is going to seek another sharing partner. Resulting in there being now one electron too many in the newly formed sharing couple. So that electron is spit back out.

Now we are going to alter this intrinsic material by shoving in impurities. called 'doping'. we can shove in a material that has spare electrons or a material that doesn't have enough electrons.

doping the intrinsic semiconductor starts an electron exchange process.
if you feed extra electrons, one of the paired atoms will absorb it temporarily causing its bonded twin to be unhappy now and seek another partner. three is company .. , also in electron land and The newly partnered atoms now have an excess electron being spat out . This process goes on and on. These 'traveling electrons' hop from bonded atom pair to bonded atom pair and are called free electrons. this is N doped material

if you feed a material with shortage of electrons ( holes ) the same happens, only now does the 'vacant spot' ( the hole) travel around. the lacking material snatches an electron from a bonded atom pair , one atom in the bond is now unhappy and goes in search for an electron it can borrow from an adjacent pair. causing a marital rift in that pair , divorce and another atom going in search for an electron ... so the 'vacant spot' also travels.

now. if we take such a doped piece of material and we shove in electrons things happen differently. i forgot the metallic - oped semiconductor process operation. it is the shottky effect ( a shottky diode is essentially a diode made from one doped material ( p material) and a simple conductor not relevant for this discussion. i;d have to read up on that again.

but , if we take a lump of P material and a lump of N material and we slam them together something happens. in the contact surface the free electrons from the N material will 'fall' in the holes of the P material making atomic bonds (the recombination zone) . since in that region there are now no mobile charge carriers, this is effectively an isolator !

now, if we apply a field to such a diode. , in the right direction , we will be able to send electrons into it , but none will come out the other side ! you will need to keep increasing the external field until you hit the point where the recombination zone is gone. essentially this zone travels in the material. add some electrons in the N material by applying  a field and they will push the recombination zone to the exit ( other end)  once you hit enough potential the recombination zone is gone and electrons flow freely. this potential is called the forward voltage. ( 0.6 volts in typicla silicon diode) and is depending on the levels in the valence bands of the used material.

now. if we make a bipolar transistor we get two such recombination zones. by applying pressure to the base-emittor (biassing) we get that thing to go in conduction. now we have established an electron flow there. ( a current). that electron flow now pushes out the second recombination zone and allows collector current to flow. ( i have described that process already once in detail here on the forum, i'm not going to repeat it here )

change the intensity of the emittor-base current and the intensity of emittor-collector current changes. ( electron model currents )

so , to put the dot's on the 'i':  for a properly biased bipolar transistor ( bipolar as it used both electrons and holes as charge carriers) the collector current has a relation to the base current. once base current flows, both recombination zones are gone and collector current flows. send in more electrons to the base and more will flow toward collector. depending on the strength of the doping there is an amplification factor.

ergo : such transistors are current controlled.

internal material resistance cause losses, so you get all kinds of side effects such as temperature dependency of certain factors . there is also charge accumulation due to uneven material properties ( microcracks in the lattice , uneven distribution of the impurieties etc ) so you get all other kinds off effects. if you flood in enough electrons to completely collapse the recombination zone you are in 'saturation' and then other effects kick in as well. these are well understood and described in various models and equations. but none of that erases the base principle :

make base current flow to get collector current. increase base current to get more collector current. so : current controlled current ampliefier.

now do you get it ?

[c4757p]

now, if we apply a field to such a diode. , in the right direction , we will be able to send electrons into it , but none will come out the other side ! you will need to keep increasing the external field until you hit the point where the recombination zone is gone. essentially this zone travels in the material. add some electrons in the N material by applying  a field and they will push the recombination zone to the exit ( other end)  once you hit enough potential the recombination zone is gone and electrons flow freely. this potential is called the forward voltage. ( 0.6 volts in typicla silicon diode) and is depending on the levels in the valence bands of the used material.

This goes completely against every explanation I've ever seen of PN junctions, where the depletion region is constricted inward from both sides, not pushed out one end... how do you plan to "send electrons in" without removing electrons from the other side at the same time?

[T3sl4co1l]

Right, if nothing's coming out the other side, you have a capacitor.  Electrons pile up until the voltage is equalized against the source.  Which is an approximate model of a diode at V < Vf (including negative, i.e. reverse bias).

Although the depletion region does shrink under forward bias, the usual story given is that the applied voltage skews the electronic band gap diagram; in particular, when Vf > E_g / q_e, the diode is in forward bias.  This causes holes to leak into the electrons region, and vice versa; electrons and holes are freed purely by thermal means (minority carriers), so there is a statistical population of charge carriers (number and energy), which is where the exponential characteristic arises from in semiconductors.

Which is why, although a depletion region can be approximated as a capacitor, it's actually a very leaky capacitor, if at all.  DC current flows across the junction, due to random carriers generated in the gap (reverse bias), or in bulk (forward bias; charge carriers move into the depletion region, helped along by the flow of current).

If depletion region thickness were the only factor, one would expect quantum tunneling to be an important part: in fact, it is not, and the depletion region is much thicker than this characteristic length in most diodes.  Diodes made with particularly strong doping (and therefore very low breakdown voltages and very thin depletion regions) do exhibit tunneling, and are called tunnel diodes.

There's also something quantum about the Zener effect, which is a low voltage breakdown phenomenon, but I don't remember offhand how it works.

Tim

[LvW]

so , to put the dot's on the 'i':  for a properly biased bipolar transistor ( bipolar as it used both electrons and holes as charge carriers) the collector current has a relation to the base current. once base current flows, both recombination zones are gone and collector current flows. send in more electrons to the base and more will flow toward collector. depending on the strength of the doping there is an amplification factor.

ergo : such transistors are current controlled.

It`s just a claim and, more than that, not logical.
Nobody has denied that there is a "relation" between both currents.
We simply have two events (currents) caused by the applied voltage. Such things happen in our world (electronic, natural) from time to time.

I have another question:
How do you explain the Early effect with your current-control view without using the electric field (voltage) causing the Ic increase?

(By the way: Did you forget to answer my technical question in post#61?)

[wazzokk]

Nobody knows at the deepest level how physics really works, all we have is various models of how materials behave.
What matters is which model is appropriate to our requirements.

Dave

not true. There exists equipment that can actually track electrons flowing in material (e-beam probing ) . IBM even has a machine that can arrange individual atoms to form the IBM logo... they have made a twenty atom transistor with it. and yes, it does work. they can track the electrons flowing. not simulated. measured.

e-beams are frequently used to sniff out leakage in integrated circuit design as you can count individual electrons.

Apologies for my sloppy description. I was trying to say that even electrons, protons, quarks etc. are still, I believe, a model for the behaviour of something, string theory maybe, we have yet to understand. And so we use models apropriate to our needs.   

Respectfully Dave

[LvW]

If I have read right people are saying that the art of electronics supports "the voltage theory" well I've just skimmed through my copy and all formula's that involve voltage also involve resistance values in a circuit!

Of course voltage has a part to play but it's not strongly linked to BJT basic behaviour other than the obvious requirement for a voltage in order for current to flow.

Hi Simon,
excuse me but I totally have forgotten to answer your request regarding Horowitz/Hill.
Here is an excerpt from chapter 2.09 "Transconductance":

Clearly our transistor model [LvW: current-control] is incomplete and needs to be modified in order to handle this circuit situation, as well as others we will talk about shortly. Our fixed-up model, which we will call the transconductance model, will be accurate enough for the remainder of the book.

And at the beginning of chapter 2.10:

But to understand diff. amplifiers, log. converters, temp. compensation, and other important applicati0ns you must think of the transistor as a transconductance device - collector current is determined by base-to-emitter voltage.

(This justifies my "stubborness" you have attributed to me.)

[free_electron]

now, if we apply a field to such a diode. , in the right direction , we will be able to send electrons into it , but none will come out the other side ! you will need to keep increasing the external field until you hit the point where the recombination zone is gone. essentially this zone travels in the material. add some electrons in the N material by applying  a field and they will push the recombination zone to the exit ( other end)  once you hit enough potential the recombination zone is gone and electrons flow freely. this potential is called the forward voltage. ( 0.6 volts in typicla silicon diode) and is depending on the levels in the valence bands of the used material.

This goes completely against every explanation I've ever seen of PN junctions, where the depletion region is constricted inward from both sides, not pushed out one end... how do you plan to "send electrons in" without removing electrons from the other side at the same time?

Don't get all upset about what i will state now, and i will go out on a limb here... Everything you know is wrong... Not your fault, nobody ever explained it to you the right way. They sketch a pciture but it is not complete. Even i don't have the full picture as i never really had to delve that deep. The people that make transistors do. I got what i know from them, some of that stuff i don't grasp but other stuff i do.

So let me try to fill in a few gaps.

A diode that is not in conduction (whether forward or reverse) is indeed a capacitor.
Remember the varactor diode ? You bias that thing in reverse . Apply more voltage and the recombination zone grows thicker , thus capacitance goes down.
A diode in forward also has capacitance. In this situation you apply a field , causing electrons to move to collapse the recombination zone. A field travels.. As charged particles . Charge carriers are electrons. Traveling charge is a current.

To get a diode into conduction you need to send some electrons in. You can actually measure that.
Take a voltage source of a few hundred millivolt, well below sending the diode in conduction. Take a very sensitive coulomb meter and you will see a current for a very short time enough to charge the barrier. It takes time for a dide to turn off as the internal charge needs to come out to allow the recombination zone to grow. Diodes have turn on and turn off times due to the capacitance !

Other things like the miller effect are similar things.

Look up 'capacitance of forward biased diode'
http://web.ewu.edu/groups/technology/Claudio/ee430/Lectures/pn_mos_2.pdf

So back to the transistor : as long as you do not get a base current flowing no collector current flows.
See what happens if you try to apply 10 volts to the base emittor of a transistor. It thermally self destructs. The 9.4 unnneeded volts cause such high pressure in the resistance causeing so many electrons to bang around that you destroy the pathway.

Bipolars are current controlled. I stand by that statement.

[c4757p]

But nothing you just said contradicted what I said!

A diode that is not in conduction (whether forward or reverse) is indeed a capacitor.
Remember the varactor diode ? You bias that thing in reverse . Apply more voltage and the recombination zone grows thicker , thus capacitance goes down.
A diode in forward also has capacitance. In this situation you apply a field , causing electrons to move to collapse the recombination zone. A field travels.. As charged particles . Charge carriers are electrons. Traveling charge is a current.

Right, obviously.

To get a diode into conduction you need to send some electrons in. You can actually measure that.
Take a voltage source of a few hundred millivolt, well below sending the diode in conduction. Take a very sensitive coulomb meter and you will see a current for a very short time enough to charge the barrier. It takes time for a dide to turn off as the internal charge needs to come out to allow the recombination zone to grow. Diodes have turn on and turn off times due to the capacitance !

Right - now put that coulomb meter on the other side. I'd bet serious amounts of money (if I had them...) that you'll see equal amounts of charge going out the other end, whether or not the diode is in conduction - hence the depletion region will be constricted inward by the symmetric change in charge, not pushed towards one end like you said.

Bipolars are current controlled. I stand by that statement.

It's a bikeshed. Most of us aren't arguing about how BJTs work, we mostly seem to understand that. Whether we realize it or not, most of us seem to be bickering about which aspect of the transistor's behavior is the control - something about which the charge carriers don't give a damn!

[free_electron]

Ah, if you want to put it that way. What is the control ?

Collapse all the equations and what do you get for a single transistor (not a system built around one) ?
Ic = ib x beta.

Now tell me what the symbol I represents.

Connect the dots.
And that's all i will say about that. This is an endless discussion leading nowhere. It's righ there in that equation.

[c4757p]

Right, and people are arguing because there are other ways to collapse those equations!

[mtdoc]

Interesting discussion and fun for an electronics newbie like me to follow.  Lots to learn from the more educated and experienced posters when they disagree.

The following seems apropos:

[Zero999]

After doing some more research, I've realised no one here is right. BJTs are neither current nor voltage controlled devices. The truth is as BJT is a temperature controlled device! 

[c4757p]

Nuh-uh, they're obviously light-controlled. What, are you people leaving them in the box?

[free_electron]

After doing some more research, I've realised no one here is right. BJTs are neither current nor voltage controlled devices. The truth is as BJT is a temperature controlled device! 

nonsense. they are smoke controlled. once the smoke escapes it's game over.  that is why we have clean rooms. we take all the smoke from the room and deposit in the transistor package. that is why that room is so clean. oops. now the dirty secret is out ... shouldn't have said that ...  the semiconductor police is now out to come and get me . they will take me a way to the funny farm where i will have tea with unicorns until eternity. damn !

[free_electron]

Right - now put that coulomb meter on the other side. I'd bet serious amounts of money (if I had them...) that you'll see equal amounts of charge going out the other end, whether or not the diode is in conduction - hence the depletion region will be constricted inward by the symmetric change in charge, not pushed towards one end like you said.

you will see a charge displacement but the current in will not be equal to current out. some electrons stay behind to bind with the atoms to erase the holes. once the thing goes into conduction electrons in will equal electrons out. turn off the diode and current flows a bit longer than the point where it stops conducting. those are the electrons now being released to reform the recombination zone.

Now, as for the 'symmetrical' that is not right . it depends where the electrons are on the shell to knock em off. or seed them in. i can't remember it exactly but the layer shrinks assymetrically. i depends on the dopant used and how strong the bonds are. where they sit in the periodic table has an impact. boron, antimony, phosphorus and arsenic are the traditional dopants for silicon or germanium processes. you need a materials chemist to tell you how that stuff happens.

and then that is only theoretical. there are flaws in the crystalline structure of the semiconductor. during implantation we shoot the cristalline structure to shreds. part of that is repaired during the drive-in process step. after implanting the wafers go in an oven. this does two things : it drives the dopant deeper in the wafer vertically and it heals the cristalline structure.
the deposition in vertical is not uniform. there are more dopants per cubic micron at the surface than 2 micron deep. so the deep regions actually kick in first as you need more charge at the surface to break down the zone there.

area wise (seen from the top ) the doping is pretty uniform as the wafer spins through the beam thousands of times when being implanted. vertically it is not.

so that barrier does not dissolve symmetrically.

on a transistor the barrier does shift. keep in mind that we are dealing with layers that are a few thousand atoms thick. i think 10.000 atoms is required for a micron. that base layer isn't a micron, it's less. and the whole recombination zone isn't the full width of the base either.

the distance between the b-e barrier and the c-b barrier is very small. the zones are built-off asymetrically. the be zone collapses first. it needs to as it hinders the c-e connection. once the b-e zone is gone electrons shoot from the emittor to the collector ( electron model) and start collapsing the recombination zone there.

so once again : it is the electrons flowing from emitter to base region that make electrons from emittor shoot into the collector.

think of the damn thing as a venturi. stop the electrons fro coming out of the base and none will come out of the collector.

so again : current controlled. no traveling electrons , no amplification.  sure you can blabber that you need a field first to make them move in the first place , but you can put a million volt
9provdied the part doesn't flash over) at the emittor , if the base does not emit a single electron you have no current there. , nothing out of the base is nothing out of the collector.

a mosfet is different : there you do build a static field in the part , that attracts charge carriers on the other side of the barrier ( the gate isolation ) and then you can send electrons through that pile of atteacted carriers . so mosfets are voltage controlled ( they have current flowing in and out changing state from field to no field and vice versa. ( on and of )

bipolars are current controlled.

[IanB]

you will see a charge displacement but the current in will not be equal to current out

It is one of the most fundamental laws of electricity that current in equals current out. The is Kirchhoff's current law. I really do not think that semiconductor devices violate this law. If you measure the currents at all the terminals of an electronic device the sum is going to be zero.

[c4757p]

To be fair, that fundamental law is more properly that current in equals current out + current stored - but I'm still not convinced. This one'll take some consideration...

I'll give him that the depletion region doesn't shrink truly symmetrically, this will obviously be subject to all sorts of variations. But it sure as hell doesn't get pushed out one end.

[free_electron]

To be fair, that fundamental law is more properly that current in equals current out + current stored - but I'm still not convinced. This one'll take some consideration...

I'll give him that the depletion region doesn't shrink truly symmetrically, this will obviously be subject to all sorts of variations. But it sure as hell doesn't get pushed out one end.

fundamentally it is electrons out = (electrons in - electrons stored). moving electrons are current. stored (non-moving) electrons are charge. potential energy <-> kinetic energy.

when i say ' pushed out ' this is in reference to the transistor. in a diode that region sits at the junction of the n and p material. thats why we call it the junction.

in a transitor the two regions sit very close together , they are really only separated by the width of the base.

so when you start deconstructing the b-e junction that barrier becomes thinner. as it shrinks asymetrically it has the perceived effect of shifting towards the base and away from the emittor. once gone, the electrons flow freely into the base region. as the base is doped more weakly than the collector , and the potential at the the collector is (typically) higher ( unless you run in inversion mode) some electrons will happily run into the base, others , which are mechanically further away from the attraction of the base potential , will shoot into the collector.

in simple words : the collector is a wider pipe which is easier to traverse than the base pipe. and the 'pull' from the potential at the collector is higher than that of the pull from the base.

so if you want to do the 'voltage controlled' thingy it would actually be the collector that has control... is it has a stronger pull.

but it all comes back to the fact that no base current is no collector current, irrespective of what voltages you put there. if you dont succeed in collapsing the b-e barrier and get a current to run there, then no current runs in the e-c path.
no current flow in b-e is no current flow in c-e.  ergo : current control.

[Syntax_Error]

This is a very interesting discussion.

free_electron, if I understand your posts correctly, the following seems a logical extension of what you are saying:

As the E field pushes charges into the diode substrate, the diode should technically become charged during forward biasing operation, i.e. not strictly neutral. Can this be measured with either an electroscope or some sort of E field meter?

[free_electron]

This is a very interesting discussion.

free_electron, if I understand your posts correctly, the following seems a logical extension of what you are saying:

As the E field pushes charges into the diode substrate, the diode should technically become charged during forward biasing operation, i.e. not strictly neutral. Can this be measured with either an electroscope or some sort of E field meter?

yes. this can be measured (i'll show how in a moment) . the diode is charged BEFORE it goes into conduction. it behaves like a capacitor until conduction. the perceived effect of this is the turn on time of a diode. you apply a field , current starts flowing in , but nothing comes out until the barrier is charged enough so it becomes conductive. then stuff comes out. that delay between current in and current out is directly proportional with the physical size of the diode. make a big bulky diode of 10 square meters and it will be slow as molasses. make it a square nanometer and it will be lightning fast. why ? capacitance ... (what we call capacitance is charge storage. we defer electron flow to a later point in time )

same goes for turning off the diode. stop cramming electrons into a diode that is in conduction. the current flows a bit longer at the output. where do those guys come from ? they are released from the junction . so the recombination happens , electrons are spat out and the barrier is recreated.

you don''t need special equipment to measure this.
two current probes , a scope and a pulse generator. look at the phase shift of the currents. if you dont have current probes : a diode followed by a resistor. measure vin and voltage across resistor. you will see the shift. That is the turn on and off time of the diode. cause by charging and discharging the barrier.

How much is that charge you ask ? measure the drop across the diode, look a the turn on time you can find out how many coulombs...  does the coin drop now ?

there are of course some non linear effects due to material physics and resistive losses and temperature making stuff jiggle faster or slower.

No laws of physics are broken as , integrated over infinite time, electrons-in = electron-out of this system. there is just a temporary hold on some of them to destroy/reconstruct the barrier.
that 'electron storage' is seen as the phase shift in current. kirchoffs laws apply only to effects integrated over time. that is why you can use them for ac signals with impednce like capacitors and inductors. look at them frozen in time and they make no sense as they do not account for stored charge. you need maxwell for that crap ...
Kirchoff only works on conducting diodes. not to look at transient behavior. ( actually it does but you need to track every damn electron as it travels through material at infinitesimal time fragments.)

i'm not pulling this stuff out of my ass you know. this is how these things work. It's all really very simply if you think about it. problem is we are always being told half-complete stories. i know we have to start somewhere , but i feel too much is left out in classical literature to explain it in enough detail so it makes sense. later on you can slap on the -very complex- mathematics.

[Syntax_Error]

I don't think anyone's accusing anybody is pulling stuff out their ass. I think the various players here all genuinely think their understanding of things is the correct version, and I think to various degrees, they may all be. This sounds quite politically correct, but I'm not trying to make it that way on purpose. I think there
s a bit of cognitive dissonance going on, and it is being handled in the most commonly human way: instant rejection of the new information that drives the dissonant feeling.

free_electron, your information regarding fields and charge accumulation make perfect sense to me. I always wondered about this, since for the life of me I have never been able to understand how voltage and charge could ever be separate, unrelated quantities, based on my understanding of their definitions. In order for a node to have a different potential than another node, it's charge *had* to be different, or it's physical location within an external electric field would have to account for all of it's voltage. It makes so much more sense to me that the voltage (in terms of P/Q) of a node is due to the presence of charge, however small. I relate the quantity of charge to the voltage related to it as the capacitance. A diode with very small capacitance would accumulate very little charge for any given voltage. A diode with large capacitance would accumulate more charge for that same voltage, and conversely less voltage for the same charge.

Edit: Not to imply that the forward voltage drop is due to accumulated charge. Rereading my post made it seem like this is what I was implying.

[IanB]

To be fair, that fundamental law is more properly that current in equals current out + current stored

Yes, I understand that, and in a monopolar device like a Van de Graaf generator there is a transient accumulation of charge. But the generator has a big metal dome and a current imbalance of a few microamps produces a voltage rise of thousands of volts.

A P-N junction is a bipolar device with two terminals, like a capacitor, and it is on a totally different microscopic scale compared to a VdG. So sure, on some scale there may be a transient charge accumulation term, but we must be talking of scales like femtoamps and picoseconds. Being a bipolar device the minutest charge imbalance will produce a current on the other terminal.

Is there a supporting reference somewhere for this?

(To give something for comparison, consider an actual, big capacitor. Feed it with an AC signal and measure the current at each terminal. Can you measure a phase difference between these currents?)

[free_electron]

A diode with very small capacitance would accumulate very little charge for any given voltage. A diode with large capacitance would accumulate more charge for that same voltage, and conversely less voltage for the same charge.

Edit: Not to imply that the forward voltage drop is due to accumulated charge. Rereading my post made it seem like this is what I was implying.

keep in mind that ,as the charge increases, the capacitance increases. in a standard capacitor this doesnt happen.
a higher voltage, but below conduction decreases the thickness of the barrier. thinner barrier = larger capacitance for the same surface area ( capacitance is inverse of distance between the plates )
so you may start of with a small capacitor but it actually increases as the barrier starts to collapse.

[T3sl4co1l]

It's worth noting that Kirchoff's law (it's not a rule, despite being labeled as such in most references) is absolute.  But if you want to niggle every little detail, you must understand the scope of that law.

- Kirchoff's is absolute at DC.

- At AC, it is instantaneous and pointwise.  Currents can only be truly said to sum at a node if that node is infinitesimal.  Alternately, if the speed of light is infinite.

- Kirchoff's is local.  If you consider a current flowing in a wire, but don't consider the transmission line nature of that wire, then you might lead yourself into a trap, where the two ends of that physical conductor apparently violate KCL.  This is incorrect.  When the wire is correctly modeled as a two port transmission line, then it becomes obvious that, at each end, you are measuring the current between the active wire and ground (or whatever it's physically closest to).  The physical mechanism is, the current is being transported as displacement current in the electromagnetic field.  Where the transmission line itself leads, is completely irrelevant -- which is one of the fantastic features of transmission lines, that they offer a path for signals to flow, independent of the environment nearby.  (The real world isn't quite so fairy-tale, but transmission line transformers can be constructed which do an excellent job of this over a wide frequency range.)

The proceeding conversations about the nature of charge and depletion regions is superfluous (you're arguing the wrong thing), because for any frequencies where we can reasonably discuss the nature of semiconductor devices (i.e., DC to ~GHz), the pointwise / infinite-speed-of-light model is sufficient.  You push current into one terminal, the same total amount necessarily must flow out the others, period, no ifs, ands or buts.

Tim

[free_electron]

i am not arguing with kirchoff. kirchoff is right , but only for static systems. you cannot use kirchoff to solve transient effects.

take three capacitors , connect them in star. two go through a resistor to ground. apply a between the third input of this network and ground. good luck with kirchoff..
accoording to kirchoff no current flows as capacitors block dc. yet electrons have moved !

the diode does not violate this . a portion of the electrons is temporarily stored and released time-shifted. the total sum of electrons in is equal to the total sum of electrons out. kirchoff was not violated. the storage is happening as electron-proton bonds in the semiconductor material when the recombination zone is deconstructed. that zone discharges them eventually when it is reformed as the diode turns off.

nothing wrong there