# Goal of Parameter Extraction

lE0 bE0 usw. einführen, skalierbares Modell. Wann sind Parameter optimal?

# HICUM/L2 Extraction Flow Outline

....

## PoA Seperation

The first step required to extract scalable HICUM/L2 model parameters is seperating the geometry specific components of the base/collector current and capacitance. For this purpose, DMT implements a Perimeter-over-Area analysis based on Bilinear Scaling Theory [@REFTODO]. PoA analysis enables to separate every electrical quantity into its emitter window area, width and length specific components.

The length and widht drawn in the layout is not exactly equal to the actual width of the fabricated emitter window. To capture this effect, the constant offset $\delta_{El}$&\delta_{Eb} are introduced, which relate the drawn widht/length to the actual quantaties:
$$
l_{E0} = l_{E,draw} + \delta_{El} \quad\quad b_{E0} = b_{E,draw} + \delta_{Eb}
$$

Furthermor corner rounding has to be taken into account for accurate scaling of state-of-the-art processes [@REFTODO]. For such purposes the radius of the rounded corners $r_{ecr}$ is also required.

While the constant offsets and the corner rounding radius can be determined using numeric techniques, ideally they are provided from TEM/SEM data.
Hence, these parameters are written explicitly into the modelcard before the PoA analysis is performed.
To enable PoA extraction the `XQPoaBilinearFull` class is provided in *DMT*

The range of operating points should be selected by the device engineer.
Generally, the PoA analysis is valid only at low injection. Hence, the device engineer needs to select an appropriate range by looking at the measurement data before configuring this extraction step [@REFTODO]. The result of an exemplary PoA seperation for $C_{BE}$ are shown in [@fig:PoA_CjC]

{#fig:PoA_CjC}

## Resistances

### internal base series resistance

* Tetrode
* Simple

### sheet resistance

### Rth

* could be done using method from [@pawlak14] but that doesn't work that well in practice (**why?**)
* fit rth by fitting $\frac{\partial V_{be}} {\partial P}$ by numerically differentiating the temperature network and the ibei,s
* only valid where derivative is roughly constants (for fairly high vbe **why**). **source for approximation**
* also fit artifical diode parameters here