Mesa Production 1958 Protective Coating Exposed Junction

# Attempt a Protective Coating Over the Exposed Junction: A Partial Answer, Honestly Measured

## 1. Why Covering the Edge Helps Without Actually Solving the Problem

This step applies a thin organic coating — a silicone varnish or a similar lacquer, dipped or sprayed onto the diced die after Step 5's etch — directly over the exposed mesa sidewall, in the hope of blocking the moisture and ionic contamination Step 6 showed drives surface leakage upward. The coating is not a barrier in the sense Step 1 of the 1957 series' thermal oxide was a barrier: it is not grown from the silicon itself, it does not bond chemically to the surface the way an oxide does, and it is applied by hand or by dip at a scale where perfect, pinhole-free coverage on every single die is not realistic. The honest way to describe what this coating buys is as a partial, imperfect reduction, not an elimination:

where $\theta$ is the fraction of a die's mesa edge that the coating actually covers without a pinhole, $D_{\text{pin}}$ the areal density of coating defects, $A_{\text{edge}}$ the exposed sidewall area, $I_{\text{bare}}$ the leakage Step 6 measured on an unprotected edge, and $I_{\text{reduced}}$ the leakage under an intact section of coating. The structure of $\theta$ is the same Poisson relationship this project used for defect yield in the 1957 series — which means the coating's effectiveness is itself subject to exactly the kind of defect-density economics that governed photolithographic yield, except here the "yield" being protected is reliability rather than a working device at all.

## 2. Real Diagram: Where the Coating Helps, and Where It Quietly Fails

A coated die looks uniformly protected from the outside. Underneath, coverage is never actually uniform — thin spots at sharp mesa corners, bubbles trapped during dip application, and cracking as the coating ages under thermal cycling all create local gaps where the bare junction is, in effect, still exposed, indistinguishable from an uncoated edge at exactly those points.

## 3. Why This Step Does Not Resemble Anything in the 1957 Oxide-Masking Series

This project already documented a far more rigorous way of covering a silicon surface: the thermal oxide the 1957 series grew in Step 1, which forms by consuming the silicon's own atoms and bonding chemically to it. That oxide's reliability as a barrier comes precisely from the fact that it is not a coating applied from outside — it is a new material grown directly out of the surface it protects. This step's varnish is everything the 1957 oxide was not: applied after the fact, bonded only by adhesion rather than chemical growth, and subject to defects that have nothing to do with the silicon underneath and everything to do with how evenly a liquid happened to flow over an irregular mesa shape during a dip. The two approaches are not different strengths of the same idea; they are different kinds of idea entirely, and only one of them, 1957's, was ever capable of the reliability this product actually needs.

Step 7 does not solve the exposed-junction problem; it buys this line a partial, quantifiable, honestly-reported reduction in how often that problem shows up, while leaving the underlying defect — an edge that was never meant to be covered by anything other than silicon's own chemistry — fully intact.