scott-maddox / obpds Goto Github PK

Add a bias control to the right hand side. Clicking on it would set the bias and update the band diagram.

The plot should not decrease the scale through autoscale.

Similar to BandProf

Medium term:

• basic finite quantum well simulation:

  QuantumWell(width=10*nm, well=GaInAs(In=0.1).strained_001(GaAs), barrier=GaAs).E(n=1)
  

  • QuantumWell would inherit from LayerStructure, so that it could
    be inserted into structures and simulated.
  • You could even include basic quantum confinement effects on the carrier
    concentration in the simulation this way, by adjusting the
    band edge potentials and effective intrinsic carrier concentration.
    Probably overkill for most things, but it could come in handy.

Prof. Toshio Fukushima has a pre-print paper on this: DOI: 10.13140/2.1.1094.6566

If it's quick enough, I can use it by default

The capacitance can be efficiently and accurately calculated using the method described in [1].

[1] R. C. Jaeger, F. H. Gaensslen, and S. E. Diehl, “An Efficient Numerical Algorithm for Simulation of MOS Capacitance,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, vol. 2, no. 2, pp. 111–116, Apr. 1983.

• implement Cache class
  • functions similar to a dictionary
  • takes a maximum size upon instantiation
  • use FIFO rule to limit size

plt.style doesn't exist on the version available in the Xubuntu 14.04 repositories.

Include band non-parabolicity effects on the Fermi-Dirac statistics

[1] V. Ariel-Altschul, E. Finkman, and G. Bahir, “Approximations for carrier density in nonparabolic semiconductors,” IEEE Transactions on Electron Devices, vol. 39, no. 6, pp. 1312–1316, Jun. 1992.

Currently, each materials parameter is stored in a separate array. Ideally, the parameters that are accessed together would be stored together, so as to minimize cache misses.

I'm not sure that this is currently the performance bottleneck, though, so some benchmarking is merited.

• Include the effects of strain on the bandgap and band offsets, using deformation potentials.
  • Currently, each layer can be specified as strained, but having a simplified way of doing this might be preferable. It should also be possible to specify that a layer is partially or fully relaxed, and the layers above it should take on the new lattice constant.
• Solve for the strain profile before simulating the band structure.
  • Discontinuities in the strain profile are only an approximation. In
    reality, the layers surrounding a highly-strained layer also have
    strain. This is critical for the band offsets of highly-strained
    quantum wells.
  • This requires significant alteration to the current (manual) way of specifying strained layers.

Test in venv

Implement Schottky contacts.

• Default the pinning location to the "universal Fermi stabilization" energy of about -4.9 eV from vacuum.
• Alter the Dirichlet boundary conditions appropriately

• move approx into it, possibly renaming it something more descriptive
• eventually add bandgap narrowing as an option, etc.

Assuming slowly varying heterjunction band offsets

Make it optional, using a boolean flag.

To handle quickly varying heterojunction band offsets