When I add the option "space charge conservation" included into the module "semiconductor", and assign a specific domain (characterized by a insulating material) to this option, I can compute the whole study. It should work, but I do not why, it finally converges to a solution where D·n = 0 at the boundary of this insulating material with a highly doped semiconductor (to a metal grade), where D·n=0 should not be a suitable boundary condition.
Generally: there is no difference between krystalline insulators and semiconductors.
But I suppose that the problem you met is due to some narrow field of vision of the used model&mathematical instument. Anyway, (in this case) in the highly doped semiconductor the donor states are fully screened and maybe this fact is taken into account in the calculation procedure. BUT: maybe!
Dear Angel Luis,
First of all I would like to know what is D.n=0, The I can give you detail explanation.
Both materials hve only completely full/empty bands at 0K. The only difference between insulators and semiconductor is the band gap energy value: if this energy is of the order of KbT at room temperature we have a semiconductor because we will have excitons present at this temperatures, if it is much larger we will have an insulator type behaviour at room temperature. This means, of course, that all insulators will behave as semiconducotrs if we heat them enough!
If you can measure reflection spectra in whole infrared region you can reconstruct the dielectric function which has two parts with phonon and plasmon parameters... For highly doped semiconductors it is important to measure at several temperatures of the sample.
Best regards, Evgeny Vinogradov
Cool it down. At temperatures near to 0K are all semiconductors dielectrics.
Dear Miloslav Frumar
Here I would like to tell you, not all the semiconductors are dielectrics, because all dielectrics are insulators but all insulators are not dielectrics
Dear colleagues,
I believe that Angel Luis knows thoroughly the physical phenomena mentioned here, his question arises because the Comsol's obtained solution apparently is not the correct one. I will study what you comment in order to give you some help later.
Regards,
Generally: there is no difference between krystalline insulators and semiconductors.
But I suppose that the problem you met is due to some narrow field of vision of the used model&mathematical instument. Anyway, (in this case) in the highly doped semiconductor the donor states are fully screened and maybe this fact is taken into account in the calculation procedure. BUT: maybe!
hello, Dear Angel Luis
Basic difference between insulator and semiconductor is band gap b/w valence band to conduction band .but w use semiconductor material as insulator keeping semiconductor in low temp.(~0 K). Highly doped semiconductor is also behave insulator at 0K temp.because no migration of carrier in this stage .
Dear all, thanks for your answers.
From the point of view of a simulation software, this is the difference between insulating and semiconductor materials:
Insulating materials are those which do not have FREE charge carriers, so their electrical behavior is just described by one parameter, dielectric permittivity. Then, if you perform STATIC calculations with a software (so you are not concerned by a frequency dependent response), you ensure easier equations, a enormous calculation simplification, as well as saving computing time. In constrast, if you allow the existence of free charge, you enter into the semiconductors domain. Then you need to resort to band theory, Fermi level position, state density, doping... what means much more complicated equations and longer computing times.
In COMSOL, when you select the Semiconductor Module, you need to provide a set of material properties in order to be managed by a large set of equations. In principle, all material domains should be semiconductors. But this module includes an option called "Charge conservation" to deal with a semiconductor material in a simplified manner, namely, as an insulator with just one electrical parameter (permittivity). You may desire this option in order to reduce computing time (usually many hours) and help convergence, as it is my case. But, if you use this option, the boundary between a semiconductor and this 'virtual' insulating material is not managed properly (I think).
Thanks again for your interest.
- Badges
- Science topic