Photolithography Oxide Masking 1957 Second Masking Level

# Repeat Masking for the Second Level: Where One Wafer Becomes Hundreds of Devices at Once

## 1. Why Every Capability Built So Far Compounds in This One Step

This step repeats Steps 3 through 13 with a second photomask carrying the emitter-window pattern, which must land inside the base regions Step 13 just diffused — and every capability this series has built up until now compounds here at once: the alignment system from Step 5 finally has real silicon features to register against, the thermal-budget accounting from Step 14 becomes load-bearing rather than theoretical, and the defect economics from Step 9 start to govern the entire wafer rather than one window. This is also the step where the historical claim this project set out to document becomes literally, physically true: batch processing of many devices per wafer. What makes that claim an economic fact rather than a slogan is a relationship between die size and cost that this project has never needed to state until a wafer could carry more than a handful of devices:

$$N_{\text{die}} \approx \frac{\pi r_{\text{wafer}}^2}{A_{\text{die}}}\,\eta_{\text{edge}}, \qquad \text{cost per good device} = \frac{C_{\text{wafer}}}{N_{\text{die}}\, Y}$$

where $\eta_{\text{edge}}$ is an edge-exclusion factor for partial die at the wafer rim, and $Y$ is the Poisson yield this series introduced at Step 9. Processing a wafer costs nearly the same whether it carries one device or a thousand, so cost per device falls roughly as $1/N_{\text{die}}$ — which means shrinking a die is, economically, equivalent to making the whole process cheaper, and every decade of the industry that follows this one will run on exactly that relation.

Cost Per Device Falls as the Wafer Fills Up processing cost barely changes; the device count it is divided by does COST PER GOOD DEVICE VERSUS DEVICES PER WAFER devices per wafer, Ndie → cost / device one device, every prior series tens of devices hundreds of devices, this step onward cost/device = Cwafer / (Ndie·Y) — the same wafer, divided by a number that just grew by hundreds every earlier series in this project sat at the far left of this curve without knowing it

## 2. Real Diagram: What a Wafer Looks Like Once Registration Has Something to Register Against

A top view of the finished wafer now shows a dense grid of complete two-level devices, each with an emitter window nested inside the base region diffused one level earlier. The margin between emitter and base edge is no longer an optical abstraction from Step 5 — it is now a real, measurable tolerance on real silicon, and it is the tightest constraint this series has produced so far.

A Wafer Full of Nested Registrations hundreds of emitter windows, each inside its own base region 1956: A FEW LARGE MESAS a handful of devices, same wafer area 1957: HUNDREDS, NESTED every blue window nested inside a yellow base margin set by step 5's error budget the visual density difference is the entire economic argument of this series

## 3. Why This Is Where All Six Prior Series Stop Being the Comparison

Every series this project has documented before this one — 1947, 1951, 1952, 1953, 1954, and 1956 — produced devices essentially one at a time, or a handful per piece of starting material, with geometry defined by physical tooling built new for each device. This step produces hundreds simultaneously, with geometry defined by a photographic pattern that replicates at no additional cost once it exists. That is the discontinuity this entire series has been assembling piece by piece since Step 1, and it arrives in the same year this project's historical note already flagged: Fairchild Semiconductor was founded in 1957 by the eight researchers who left Shockley, and Philco's micro-alloy diffused transistor was bridging the alloy and diffusion eras at the same moment — the industry was reorganizing itself around exactly the capability this step demonstrates, while this step was still being worked out in the laboratory.

Step 15 does not add a fourteenth technique to this series; it is the step where everything the previous fourteen made possible stops being a capability and starts being an economy.

Take photolithography oxide masking 1957 second masking level 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