What's wrong with my ElectrostaticDifferencePotential result?

[lknife]

Dear experts,

I tried to calculated the ElectrostaticDifferencePotential of a device using the attached .py file. And the attached .png file shows the 1D plot of the potential along c direction. It seems that the length of the screening region is not long enough. My questions are, what's the y coordinate of this plot? why the value of y coordinate is so large?

Thanks a lot for anyone willing to help me!

[Anders Blom]

Because you show the Sum not the Average

[lknife]

Thank you for your kind help! I got it.

Another question, since the bias between two electrodes is (0.1V, 0V), why the potential cannot converge to 0V at the right electrode (please see the attached figure)? Is it just because of the insufficient length of the screening region?

[lknife]

What it would be if I calculate the IV curve of a device with insufficient length of screening region? Can I get reliable result of the IV curve? Since for semiconductors, the length of the screening region could be relatively long (e.g. a doping of 10¹⁹ cm^-3 corresponds to a depletion layer length of around 100 Å), thus, can ATK simulate a metal-semiconductor interface with the length of semiconductor part shorter than that?

For a realistic device constructed from metal-semiconductor interface, such as 1nm silver thin film on 2 nm silicon with the doping lever mentioned above (far less than the required screening region length of 100 Å) thin film, what its IV curve would be? How to simulate it?

[Daniele Stradi]

- What it would be if I calculate the IV curve of a device with insufficient length of screening region?
Most likely, the calculation will not converge.

- Since for semiconductors, the length of the screening region could be relatively long (e.g. a doping of 1019 cm-3 corresponds to a depletion layer length of around 100 Å), thus, can ATK simulate a metal-semiconductor interface with the length of semiconductor part shorter than that?
No. Again, most likely, the calculation will not converge.

- For a realistic device constructed from metal-semiconductor interface, such as 1nm silver thin film on 2 nm silicon with the doping lever mentioned above (far less than the required screening region length of 100 Å) thin film, what its IV curve would be? How to simulate it?
As I mentioned, the screening region must be fully included in the calculation to ensure convergence, because the potential must be flat at the boundary with the electrodes, see Phys. Rev. B 93, 155302 (2016)

https://journals.aps.org/prb/abstract/10.1103/PhysRevB.93.155302

[lknife]

Thank you very much for your reply! I have read the tutorial which is based on this paper before.

That means, ATK cannot simulate the device with very thin semiconductor films along the thickness direction. Is it right?

What's the difference between devices with very thin semiconductor film and a molecular device? If the very thin semiconductor films (e.g. 1nm) is sandwiched by two metal electrode, can ATK simulate the I-V curve for this device?

I saw there are some ATK tutorials about the molecular devices. Why the I-V for a molecular device can be calculated bu ATK without considering the length of the screening region?

[Anders Blom]

Thank you very much for your reply! I have read the tutorial which is based on this paper before.

That means, ATK cannot simulate the device with very thin semiconductor films along the thickness direction. Is it right?

What's the difference between devices with very thin semiconductor film and a molecular device? If the very thin semiconductor films (e.g. 1nm) is sandwiched by two metal electrode, can ATK simulate the I-V curve for this device?

Sure, but that is not a metal-semiconductor interface, it's a metal-semi-metal system, very different.

I saw there are some ATK tutorials about the molecular devices. Why the I-V for a molecular device can be calculated bu ATK without considering the length of the screening region?

Screening must still be considered, but in both these cases, the screening takes place in the metal, on both sides.

[lknife]

Thank you very much for your reply!

Then, what's the difference between metal-semi interface and metal-semi-metal system? what's the "interface" mean? For the metal-semi-metal system, aren't there two metal-semi interfaces, although the interface might be very sharp? I am confused.

[Anders Blom]

Metal-semi is meeeeeeeeeeeeeeeeeeeeeeeeeeetal|seeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeemi (semi-infinite on both sides)
Metal-semi-metal is meeeeeeeeeeeeeeeeeeeeeeeeeeetal|semi|meeeeeeeeeeeeeeeeeeeeeeeeeeetal (only metal is semi-infinite)

[lknife]

Thank you very much for you reply!

What I want to ask is: for the Metal-semi-metal system, do I need to consider the length (thickness) of the semiconductor part in terms of the length of screening region? For the I-V calculation of such a metal-semi-metal system, will the calculation converge when the sandwiched semiconductor film is very thin? If the calculation can be converged, since the screening length for metal is far less than that for semiconductors, we can do the calculation with shorter devices, thus save the computational time.

Do you think the metal-semi interface in the Metal-semi system is different with that in the metal-semi-metal system? e.g. for realistic Ag/Si and Ag/Si/Ag system, what's the difference between the Ag-Si interface in both systems when the thickness of Si is 2nm, far less than the screening length needed for Si?

Thank you again for you time and kind help!

[Anders Blom]

Sure, you can definitely simulate a thin semi film between metals, and as you say the device can then be shorter. But Ag-Si and Ag-Si-Ag are two completely different systems, physically, so it's not a way to "save computational time", it's simply a study of two different things. The thickness of the semi in metal-semi-metal should not be chosen in relation to screening, but to the situation you wish to simulate.

[lknife]

Thank you for your kind reply!

I have another question about the ElectrostaticDifferencePotential. The bias voltage between the two electrodes is (0.1V, 0V). I saw the 1D plot of the ElectrostaticDifferencePotential along the c axis began at -0.1V at the left electrode, while the potential did not stop at 0V at the right electrode.
Attached are the .py file for the calculation and the 1D plot of the ElectrostaticDifferencePotential.Could you please help me with this problem?


Thank you very much for your time and kind help!

[lknife]

Can anybody here help me with this problem? I appreciate your kind help!

[lknife]

Since the left part of the interface is semiconductor, I doped it with a n doping level of 5*10^19. After that, the EDP of the device is like the attached figure.

The potential is more flat near the left electrode due to doping and end at the left electrode at -0.1V. It does not end at the right electrode at 0V.