Planar Process 1959 Diffuse Under Continuous Oxide

# Diffuse Both Junctions Under a Continuous Oxide, Mesa Etch Removed From the Flow: The Step That Isn't There Anymore

## 1. Why the Biggest Change in This Series Is a Step This Series Refuses to Take

This step diffuses the base and emitter exactly as the 1954, 1956, and 1958 series already diffused them — same furnaces, same dopants, same oxide mask from 1957 governing where each diffusion is allowed to reach the silicon — and then simply stops, leaving the masking oxide sitting over both completed junctions instead of stripping it and cutting a mesa down through them the way every prior series in this project did next. Nothing about the diffusion physics changes. What changes is a decision Jean Hoerni made at Fairchild in 1959, after watching 1958's mesa production line spend nine of its fourteen steps fighting a problem the mesa etch itself created: if the oxide already covers the junction well enough to mask a diffusion, and if the interface between silicon dioxide and silicon carries vastly fewer electrically active defects than silicon exposed directly to open air, then the obvious fix for an exposed junction is to never expose it in the first place.

$$D_{\text{it}}(\text{SiO}_2\text{-Si interface}) \ll D_{\text{it}}(\text{bare Si, exposed to air})$$

where $D_{\text{it}}$ is the areal density of electrically active interface trap states, the same quantity that governed the surface leakage current the 1958 series spent Step 6 through Step 10 fighting. The oxide-silicon interface, grown directly from the silicon's own chemistry, carries orders of magnitude fewer of these states than a mesa sidewall freshly cut and left open to the atmosphere — which means the entire exposed-junction crisis this project has been tracking since 1958's Step 5 has a one-sentence resolution: stop cutting through the one material that was already protecting the junction.

The Missing Step Is the Whole Idea 1958's flow versus 1959's, at the exact point they diverge 1958: AFTER DIFFUSION mesa etch cuts here next oxide stripped, junction about to be exposed 1959: AFTER DIFFUSION oxide left in place, nothing next; the junction stays covered Dit(oxide-Si) ≪ Dit(bare Si, air) — the same inequality 1957's step 1 used, now applied to never letting go the fix for 1958's crisis is not a new material; it is simply not performing one specific step

## 2. Real Diagram: The Junction That 1958 Would Have Cut Right Through

The cross-section after this step looks, at first glance, almost identical to the state 1958's own Step 5 found itself in just before the mesa etch — a diffused junction sitting under an intact oxide. The difference is entirely in what happens next: nothing. No etch, no mesa, no cut sidewall for the rest of this series to defend.

A Junction That Was Never Asked to Stand Alone continuous oxide, uninterrupted, over both diffused regions continuous masking oxide, untouched since diffusion base diffusion emitter diffusion neither junction edge is exposed anywhere along this cross-section this picture is the entire argument of the series that follows it

## 3. Why This Step Is Defined Entirely by What It Refuses to Repeat From 1956 and 1958

The 1956 drift-transistor series and the 1958 mesa-production series both reached exactly this same structural point — a diffused junction under an intact oxide — and both then performed a mesa etch, because isolating one device from its neighbors by cutting away the silicon around it was the only isolation strategy either process had. This step reaches the identical structural point and stops, because the problem that etch solved — separating devices — is deferred to a later step in this series that will solve it without ever touching the junction at all. Every reliability cost the 1958 series paid, from Step 6's leakage characterization through Step 10's hermetic seal, existed only because that mesa etch happened. This step's entire contribution to the project is refusing to let that etch happen here.

Step 1 does not do anything the 1954, 1956, and 1958 series did not already do; it is the first step in this project's history whose significance lies entirely in the step it does not take next.

Take planar process 1959 diffuse under continuous oxide further

Ask the copilot about this term, or have our engineers assess it against your process.

Ask CFSGPT about this → Request a technical assessment → Open the CFS app