Question about band structure of SiO2 in DEVICE

device

#1

Hello everyone, I’m doing some simulations for nanowires using DEVICE. In the nanowire, there is a section which has a structure like Semiconductor-SiO2-Semiconductor. I was trying to look at the band structure, charge distribution and electric field profile at this section. But it is really confusing to me.
Firstly, the band structure doesn’t show the existence of the large band gap SiO2. I check the material database, dielectric constant is the only parameter insulators have (no band gap information). Maybe this is the reason. But won’t this affect the simulation results?


Also, when I look at the charge monitor. It shows that in the SiO2 layer p and n are not zero. Shouldn’t the SiO2 have not charge at all?

Please help me on this. Many thanks.


#2

Hi @mengx1, Sorry for the delayed reply. The reason you are getting unexpected results in both cases (bandstructure, charge) is because the CHARGE solver does not solve for electrical transport in insulators. It therefore does not calculate the band profile and charge density in insulators as well. If you plot the charge and band profile of your semiconductor-insulator-semiconductor structure, the solver simply adds the two end points in the semiconductor regions since it does not have any data inside the oxide. This is why it looks like the SiO2 region has a varying bandgap or that there is charge inside the insulator.


#3

Hi aalam,

Many thanks for the explanation. Now I understood why the plots show the results like that.

Since CHARGE cannot calculate the charge in the insulator, does this mean the calculated electric field profile won’t be accurate? dE/dx increases with N*q. But from the simulated electric field profile, it looks reasonable.


Similarly, if CHARGE assumes there are charges in insulator, will it correctly simulate MOS capacitor for instance? Thanks.

Regards
Xiao


#4

Hi @mengx1, Yes. The electric field profile that you get inside the oxide is accurate. The reason behind that is that the insulators do get included in the solution to the Poisson’s equation. This is why the dc permittivity of the insulators are required. For the same reason you will be able to simulate a MOS capacitor using DEVICE. The solution to Poisson’s equation will give you the electric field profile inside the oxide and the semiconductor from which you can get an accurate picture of the charge distribution inside the semiconductor. We do have an example of a MOS capacitor simulation using the CHARGE solver in our knowledge base (https://kb.lumerical.com/en/index.html?other_application_mos-capacitor.html). Please let me know if you have any further questions.


#5

Hi aalam,

Many thanks for your detailed explanation.

One more question is that the electric field profile shows a sharp increase in the SiO2/p-type interface. The slope is nearly as high as the SiO2/n-type interface. This doesn’t make sense to me since n>>p as shown in the charge graph. The only possible reason I can think of is the interface-trapped charges. But I’m not sure about it. Please help. Thanks.

Regards
Xiao


#6

Hi @mengx1. Can you please share a screenshot of the two electric fields? One think to note here would be that since in a MOS structure one of the metal contacts is in contact with an insulator and not a semiconductor, you will have to disable the default “force ohmic” option for that contact.


#7

Hi aalam,

The electric field profile is shown in the third post above. The structure I simulated is n-GaAs/SiO2/p-InGaAs. n-doping is 5e18 and p-doping is 5e16. Since n is much higher than p, I expect the slop of the electric field at p-InGaAs side to be much lower than the n-GaAs side. But the simulated electric field profile has a very sharp increase in the p-InGaAs/SiO2 junction. I don’t know what caused this.

I tried to change the force ohmic, it changed the electric field at metal-semiconductor interface but not the semiconductor-insulator junction. Please help, thanks.

Regards
Xiao


#8

Hi @mengx1, there should be a discontinuity in the electric field at the interface between the semiconductor and insulator. I think this is again a matter of the visualized connecting the two points on both sides of this interface which makes it look like the electric field inside the semiconductor is as high as the insulator. You can use the “arrow” mouse mode in the visualizer to look at the data points in your electric field plot. My guess would be that for points inside the semiconductors you will see that the electric field is smaller than those in the insulator. You can try to use a finer mesh at the interface to verify this. As you increase the number of points you should see that the sharp transition in the electric field at the interface always happens between the two consecutive points right at the interface confirming that it is simply the visualizer connecting those two points.