Double Diffused Silicon Mesa 1956 Sidewall Passivation
# Sidewall Passivation: Covering the One Place Where Both Junctions Touch Open Air
## 1. Why an Exposed Junction Edge Leaks More Than a Buried One Ever Would
The mesa etch and the inspection that followed it left both junctions exposed directly to the ambient atmosphere at the mesa sidewall, and that exposure is not a cosmetic detail — a junction edge sitting in open air collects surface states, adsorbed moisture, and ionic contamination far more readily than the same junction would if it stayed buried under its original bulk silicon, and every one of those surface states acts as a generation-recombination site that adds directly to reverse leakage current. This step applies a passivating coating — a thin oxide or an equivalent protective layer grown or deposited specifically over the newly exposed trench and sidewall — to restore something closer to the electrically quiet condition the junction had before the etch ever cut through it. The leakage current this step is trying to suppress adds on top of the leakage inherent to the bulk junction itself:
where $I_{\text{leak,bulk}}$ is the leakage the junction would have even if perfectly buried, $J_{ss}$ is the surface-state leakage current density at the exposed sidewall, and $P_j$ is the perimeter length of junction actually exposed at that sidewall. An unpassivated mesa adds a term to total leakage that scales with the mesa's own perimeter — meaning smaller, more numerous mesas are specifically more vulnerable to this failure mode than fewer, larger ones would be.
## 2. Real Diagram: Passivation Applied Only Where the Etch Left Silicon Exposed
The passivation step has to coat the trench floor and the mesa sidewall without disturbing the top surface of the mesa itself, where contact metallization is still to come in later steps. Selectivity here matters: this coating protects exactly the surfaces the etch created and nowhere else.
## 3. Why This Step's Failure Mode Is Harder to Catch Than Most Others in This Process
The 1954 diffused-base germanium process this project has already documented also faced exposed-junction leakage after its own mesa-style isolation, and passivation or an equivalent sealing step addressed it there as well. The difference here is diagnostic, not chemical: because this device's drift field makes its base-collector junction region electrically active in a way a uniformly doped base's junction is not, surface leakage at an unpassivated sidewall can mask or distort measurements of the drift field's own effect on device speed, making a passivation failure look, in a quick electrical test, like a weak or absent drift field rather than what it actually is — an unrelated leakage path dominating the measurement. Catching that distinction requires exactly the kind of structural inspection Step 10 performed immediately before this one; without it, this step's success or failure would otherwise only become visible much later, in final testing, where its root cause would be far harder to trace back to.
Step 11 does not change anything about the junction itself; it seals off the one path by which the outside world could otherwise interfere with how that junction behaves.