The Carrier Calculations menu decides which carriers take part in the solve. Three of the four settings are simple; the fourth exists for a specific and rather subtle problem.
01default
Both electrons and holes, everywhere, with the populations that the bands and the Fermi level imply. This is the right setting for almost everything, and it is where you should start.
02no holes / no electrons
Omit that carrier from the calculation entirely. These are blunt instruments, useful mainly for isolating what one carrier is contributing, or for getting a stubborn structure to converge so you can see what it is doing.
Treat a result from either as a diagnostic rather than a prediction. Removing a carrier that a real device has will change the electrostatics, not just tidy the plot.
03freeze minority
Both carriers are present, but in each doped layer the minority carrier is held fixed — holes in an n-type layer, electrons in a p-type one. This is the high-frequency, or isolated-region, limit.
It exists because of a real weakness in any equilibrium model. A single Fermi level fills states by energy, without regard to whether carriers can actually get there. So if a region is separated from its natural reservoir by a barrier that carriers cannot realistically cross, the model will still fill it — happily populating a pocket from a buried layer far away, through a barrier that in the real device blocks them completely.
Freezing the minority carrier stops that. The classic case is an interface hole gas that would otherwise fill from a buried p-type layer through a barrier.
Do not leave this on out of habit. It suppresses minority injection, so anywhere that injection is real — a forward-biased junction, most obviously — the answer will be wrong in a way that looks perfectly reasonable. Turn it on for a specific reason, and turn it off again.
04What freeze minority does not cover
The setting acts on doped layers only. It identifies the minority carrier by comparing donor and acceptor concentrations, so a layer with neither — an undoped spacer or barrier — has no minority carrier to identify, and is left untouched.
That matters if your blocking barrier is itself undoped, which is common. Freezing will clean up the doped regions either side and leave the barrier populated. If you need the barrier covered too, give it a light doping so the setting can see which carrier is which.
05How this relates to biased contacts
These two mechanisms solve the same class of problem from opposite ends, and it is worth keeping them straight:
- Automatic — when a contact is the only possible source of a carrier, Poisson works that out for itself and needs no setting. That case is described in Contacts Under Bias.
- Manual — when a reservoir genuinely exists but a barrier stops carriers reaching it, no amount of inspection of the structure can tell the solver that; whether the barrier blocks depends on timescales the model does not have. That is the judgement freeze minority lets you supply.
06C–V sweeps
A capacitance–voltage sweep already applies the same minority freeze internally, because that is what the high-frequency measurement it models actually does. Selecting freeze minority as well changes nothing for a C–V run.
Software Utilities 