The NP transistor
A discussion about the possibility to make NP transistors.
A bipolar transistor contains two diodes: one between base and emitter
and one between base and collector.
The reason why the diode between base and collector is needed is
evident: when an electric tension is connected between emitter and
collector, it is that diode that will block the current. (Then, when a
voltage of 0.7 V is connected between base and emitter and a weak
current passes trough, that will allow a big current to cross the
blocking diode.)
Okay, but the diode between emitter and base, what is it used for?
If it was possible to remove it, it would be an improvement. The fact
the base-emitter voltage must rise up to 0.7 V in order to allow
the weak current to pass trough the diode does not rejoice the
conceivers of electronic circuits: it often obliges to use more
electronic components around the transistor and decreases the
performances. What's more, it makes the transistor heat and prevents it
from letting the current pass in the other direction.
I think that diode between base and emitter is not necessary for the
functioning of the transistor. It is probably there for technological
reasons; because it is easier to manufacture transistors that way.
The semiconducting n+ layer that the emitter is, may be removed and
be replaced by a direct metallic contact on the p layer of the base.
That way you've no more diode between base and emitter.
That metallic contact must obey following conditions:
- The Fermi level of the metal must be slightly less than that of
the p semiconductor of the base. Otherwise it will create a diode, just
like the n+ semiconductor did. (That's the principle of a Schottky
diode: it is made out of a metal-semiconductor contact)
- The metal must be put against the p layer without metallic atoms
penetrating the p layer. Indeed, they would dope the layer heavily
positively. Result: no more electron will be able to enter the p layer,
because it would first recombinate with one of the many holes created
by that local excessive doping. There would of course be an electric
current, but made exclusively out of holes (that will not be attracted
by the collector, so there will be no transistor effect). One solution
could be to use the tunnel effect: a very thin
insulating layer that keeps base and metal apart while allowing
electrons to cross. Yet such a system has a short lifetime because of
defects inside the insulating layer: electric sparks do apear that
destroy the layer. A solution could be to use a vacuum instead of an
insulating layer.
Transistors without that diode would have serious advantages:
- They would be real symmetrical components of the field effect
transistors. Just like the gate current is null inside a field effect
transistor (at 0 Hz), the base voltage would be null inside those
bipolar transistors.
- It would be possible to manufacture transistors where base and
emitter could be swapped. Provided both are build like suggested for
the emitter. They would be able to command a continuous current with a
high frequency signal.
- The functioning at high frequencies is not hampered, provided the
metallic contact on the base passes trough a heavily positively doped
layer. Those transistors should allow a functioning at higher
frequencies, with better amplification coefficients.
(A "funny" diode is possible: the semiconductor does not heat but the
electric wire does, a few centimetres further. That's good for high
frequencies and strong currents.)