Mesa Transistor 1958 Combine Diffusion Photolithography Production

# Combine Mesa Diffusion with Photolithographic Masking for Volume Production: Two Separate Threads Become One Line

## 1. Why Merging Two Working Processes Is Harder Than Either Process Alone

This step takes the double-diffused mesa process this project documented in 1956 and the photolithographic oxide-masking process documented in 1957 — each already proven on its own — and runs them as a single continuous commercial production line at Fairchild Semiconductor, built to output the 2N696 and 2N697 switching transistors in volume rather than to demonstrate either technique individually. Nothing about the physics of diffusion or photolithography changes by combining them; what changes is that every imperfection in every sub-process now compounds across the whole line rather than being judged on its own. A production line's overall yield is not the yield of its best step or its worst step — it is the product of every step's yield, multiplied together:

$$Y_{\text{total}} = \prod_{i=1}^{N} Y_i$$

where $Y_i$ is the yield of the $i$-th process operation in the combined flow. Because $Y_i$ is always somewhat less than one, and because this merged line now has more steps than either the 1956 mesa process or the 1957 photolithography process had on its own, $Y_{\text{total}}$ falls faster than intuition about any single step would suggest — a fact that will force this series to care about process control in a way neither predecessor series ever had to.

Two Proven Processes Merge Into One Line 1956's mesa diffusion and 1957's photolithography, running end to end THE COMBINED FLOW, 1958 1956: DOUBLE-DIFFUSED MESA base & emitter diffusion, mesa etch, isolation by subtraction 1957: PHOTOLITHOGRAPHY oxide mask, photoresist pattern, selective windows for diffusion 1958: ONE CONTINUOUS LINE 2N696 / 2N697, shipped at volume Ytotal = Π Yi — every one of these boxes now multiplies its imperfection into every other neither 1956 nor 1957 had to think this way, because neither ran this many steps as one line

## 2. Real Diagram: A Boat of Wafers, Not a Single Demonstration Wafer

The 1956 and 1957 series both, implicitly, told their story one wafer at a time. Production at Fairchild runs many wafers together through the same furnace boat and the same photolithography bench, so that a single bad diffusion schedule or a single misaligned mask touches every wafer in that batch at once — the stakes of a single process error just scaled up by the size of the boat.

A Furnace Boat Carries Many Wafers at Once one bad schedule now touches every wafer in the boat, not one QUARTZ BOAT, EIGHT WAFERS, ONE FURNACE CYCLE quartz boat eight wafers, same furnace atmosphere, same temperature profile, same schedule a single out-of-spec furnace run now costs eight wafers, not one

## 3. Why Neither 1956 Nor 1957 Faced This Multiplication Problem

The 1956 series documented the double-diffused mesa process step by step, and the 1957 series documented photolithographic oxide masking step by step, but each series implicitly treated its own sequence as the whole story — there was no earlier series whose process this process had to be stitched onto. This step is the first in this project's history whose entire subject is the seam between two already-complete processes, and the seam turns out to matter: a line built from two individually reliable processes is not automatically a reliable line, because reliability compounds multiplicatively across every step, old and new alike. Fairchild's achievement here is not a new technique; it is the discipline of making the compound of two known techniques behave as well, at volume, as either behaved alone.

Step 1 does not invent anything this project has not already documented; it is the step where 1956's process and 1957's process stop being two separate chapters and become, for the first time, one single thing a customer can actually order.

Take mesa transistor 1958 combine diffusion photolithography production 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