Si Bandgap in Device Modeling (Silvaco Atlas)

[cmos]

I recently noticed that in Silvaco Atlas (device modeling software), the default bandgap of silicon is given as 1.08 eV. This is at 300 K, no bandgap narrowing, and no other secondary effects - it is default. However, most (all?) of the modern references (e.g. Sze, Streetman, Pierret) give the silicon bandgap to be 1.12 eV.

Going back to the Atlas manual, it says the following, "... although these band parameters may be physically inaccurate compared to bulk silicon measurements, most other material parameters and models are empirically tuned using these band parameters."

I'm just wondering if anybody has any insight on the history of why 1.08 eV appears here or if anybody knows of other software models where this number appears.

[jsgruszynski]

The biggest reason: doping. The dopant energy levels are inside the intrinsic band-gap and thus the effective doped band-gap is smaller than the bulk intrinsic (undoped) band-gap. E.g. Boron doping shifts by 0.045 eV which is about the same difference.