TCAD Optimization of Dark Current .1 Alterations in thickness .1 Alterations in thickness

This section describe show we alter the base structure of germanium photodiode obtained in last section, and the results of how is the dark current affected by the modification. Note that in following sections including this one, simulations will be only performed on the base structure with exponential defect profile, which is more physically sensible.

The terms that are frequently used are in table 2‐1 and Fig. 2‐4. (Show again for convenience).

Terms Meanings

hGe Total thickness of Germanium diode (=hov+hBody+hRec)

hBody Thickness of Germanium body

hov Thickness of overlap

hRec Recession depth (= Si etched‐back thickness)

wGe Width of Germanium diode

Wi Doping separation; Width of intrinsic region

θ Tapered angle formed when silicon etch‐back process

tSi Thickness of silicon

Table 2‐1 Structural parameter of GePD. 

Fig . 2‐4 Structure of germanium photodiode. 

Fig. 3‐5 Simulated dark current as a function of Total Ge thickness. 

In the simulation of defects uniformly distributed in Ge layer with constant density, a linear dependence between dark current at ‐0.52 volt and total Ge thickness is seen.

This is because with thicker Ge layer, the number of uniformly distributed defects also increases, which results in more carriers generated through defects.

First, hBody, hov and hRec are altered, while holding doping energy, doping concentration and doping separation (Wi) fixed. Results show in Fig. 3‐5. In Fig. 3‐5, we can see that for both constant and exponential case, dark current merely changed by a little amount as hBody, hov and hRec are altered; there are no huge difference as long as the total thickness of germanium is the same, i.e. hGe=hRec+hBody+hov is the same. In other words, what really affects dark current is the total thickness of germanium, not the three individual parameters.

Besides, the slope of exponential defect profile data‐set is apparently smaller than the slope constant defect profile; however, two of them increase as germanium gets thicker. This is can be explained by that with larger thickness, the number of traps increases since the volume of germanium increases, so there are more SRH generation happen in that diode.

As for the reason that current grows slower in the diode with exponential defect profile, it can be seen in Fig. 3‐6, which shows that trap number is less in exponential density profile as hGe is over 0.6 µm.

Fig. 3‐6 Trap number comparison between constant and exponential defects. 

3.3.2 Alterations in width

As to alterations in width parameters, the effects of width of germanium body (wGe) and doping separation (Wi) on dark current are discussed in this section.

In Fig. 3‐7, it shows effects of width of germanium on dark currents in devices with three different thicknesses, while holding Wi and other parameters fixed. A trend can be seen that dark current is lower as wGe is at around 3 µm to 4 µm for each thickness.

Fig. 3‐7 Effects of the width of germanium on dark current. 

Fig. 3‐8 shows the behavior of dark current corresponding to differing doping separations (Wi). Dark current increases as doping separation gets wider, and the change is much larger than the effect of thickness or width of germanium (wGe).Similar trends show in samples with different thickness. Note that x‐axis of Fig.

3‐8 is the ratio of Wi over wGe, so that effects of wGe is excluded.

Fig. 3‐8Dark current of different diodes versus doping separation ratio. 

For a photodiode with fixed wGe, when Wi gets wider, traps in intrinsic region are uncovered by doping regions where carrier concentrations are very high. Although the peak value of electric field decreases as intrinsic region gets wider under same bias (see Fig. 3‐9), dark current still ramps up as long as more and more traps are exposed out and located at a rather high field region.

To conclude, the location of traps is critical. When the number of traps that located at large electric field region is increased, not only the generation sites are increased, the generation process becomes more severely as well.

Fig. 3‐9 Electric field distribution in diodes with different doping separation. 

3.3.3 Alterations in implant energy

Implant energy determines junction depth. Fig. 3‐10shows dark current of germanium photodiode as a function of implant energy (presented in E/E0 ratio, where E0 is the original implant energy). Dark current becomes higher when junctions become shallower. In Fig. 3‐11 it can be seen that as junction depth decreases, electric field spreads out more such that it promotes SRH generation process.

Fig. 3‐10 Dark current versus implant energy ratio. 

Fig. 3‐11 Electric field distribution of diodes with different junction depth 

. 

3.4 Conclusion

In this chapter, TCAD simulation is involved. In the beginning, models used in our simulation are introduced, including SRH model. Then base structures with estimated defect density are built via TCAD sprocess tool; the estimated defect densities are 5e18/cm³ and 1e20/cm³ for simulations of constant defect profile and exponential defect profile respectively. Finally we make alteration in the base structure and show an optimum structure that has minimum dark current.

Generally speaking, to make a germanium photodiode with low leakage, a thin germanium layer is required to reduce the SRH volume (number of traps) as much as possible, a narrow doping separation is crucial to keep a small SRH volume within this large electric field region, and finally deep junctions are needed to circumvent the spreading electric field.

Chapter 4 Quantum Efficiency