Tips and Pitfalls

Reference

A short list of the things that might mislead — results that look wrong but aren’t, and results that look right but are. Each points back to the page that covers it in full.

01Reading the results honestly

• A logarithmic density axis exaggerates tiny peaks. On a log scale a minority-carrier feature — an interface hole gas, say — can look like a substantial carrier gas while holding almost nothing. The density axis defaults to linear for exactly this reason. Switch to log to see a depleted region and an accumulation peak together, but read the magnitudes, not the shapes. See The Plot Window.
• The two-dimensional sheet density is the quantised one. The carrier densities on the plot and the sheet densities in the transport tab are semiclassical, with no confinement. A real 2DEG or quantum-well sheet density is the confined figure from Find Quantised States, and it is generally lower — even with quantum feedback on. See Sheet Resistance and Transport.

02Getting a solution to behave

• A biased run needs the right pair of contacts. This is a zero-current model, so a contact-bias run represents a well-posed problem only for an n-type/p-type junction, a gated structure, or an undoped stack tilted by a field. Two same-type contacts with carriers between them is a resistor, and the Biased contacts need a junction warning is the model being honest, not failing. See Contacts Under Bias.
• Freeze minority is a scalpel, not a default. Turn it on only for a region genuinely cut off from its reservoir by a barrier. Left on out of habit, it suppresses real minority injection and returns an answer that looks perfectly reasonable and is wrong. It also acts on doped layers only, so an undoped barrier needs a light doping to be covered. See Carrier Calculations.
• Heavily doped layers read low unless you tick Fully ionised. The partial-ionisation model has no impurity-band merging, so it under-predicts ionisation at degenerate doping. For heavily doped caps and contacts, tick FullyIonised so the layer carries the carriers you intended. See Temperature and Ionisation.

03Habits that save time

• When a result looks off, refine the mesh. A coarse mesh can miss a layer boundary and under-count the sheet densities. Refining is cheap — the runtime is roughly linear in the number of nodes — so drop the spacing and re-run before you doubt the physics. See Mesh Spacing.
• To see the band diagram, run without C–V first. A C–V run does not draw the bands, because the plot would show only the first biased sweep point. Run once with CV calculation unticked to see the equilibrium picture, then enable it for the sweep. See Capacitance–Voltage.
• Edit materials in the external file, then restart. The editable materials database lives outside the application, reached through Reveal Materials File; it is read at startup, so a change only takes effect after you restart. Do not edit anything inside the installed application itself. See The Materials Database.