Photolithography Oxide Masking 1957 Diffuse Through Windows

# Diffuse Dopant Through the Opened Windows: The Moment Selective Doping Actually Happens

## 1. Why the Furnace Itself Changed Nothing, and Everything Changed Anyway

This step places the wafer into a diffusion furnace using the same dopant chemistry and the same temperature schedules the 1954 and 1956 processes already used, and the furnace does not know or care that anything about this wafer is different — the only thing that has changed is what the wafer looked like when it went in. Dopant now enters silicon only through the windows Steps 10 through 12 opened and verified, and is held out everywhere the oxide still stands. For the first time in this project's documented history, one furnace cycle produces a *patterned* doped region instead of a blanket layer across the whole surface — and it does this with a vertical profile identical to every diffusion this project has already modeled, plus one new term this process has never needed before:

$$N(x,y,t) = N_s\,\text{erfc}\!\left(\frac{x}{2\sqrt{Dt}}\right), \qquad y_{\text{lateral}} \approx 0.8\,x_j$$

where the first expression is the familiar vertical dopant profile from the earlier diffused-base and drift-transistor series, and the second states the genuinely new fact: dopant diffuses sideways under the oxide edge by roughly eighty percent of the junction depth $x_j$. The diffused region is always wider than the window that admitted it, which means a mask designer must now budget for lateral spread the way Step 10 already had to budget for lateral undercut — a constraint that simply did not exist when diffusions were unpatterned.

Dopant Spreads Down and Sideways, Not Down Alone iso-concentration contours beneath one window CROSS-SECTION BENEATH ONE OXIDE WINDOW silicon, undoped outside the contours window width xj, junction depth 0.8xj, lateral N(x,y,t) = Ns·erfc(x/2√Dt), ylateral ≈ 0.8xj the dopant physics did not change; only where it is allowed to act did

## 2. Real Diagram: The Whole Argument in One Comparison

Every prior series this project has documented doped the wafer's entire exposed surface in one furnace cycle, with device geometry imposed afterward by removing material. This step, for the first time, produces several separately doped pockets on the same wafer in a single cycle, with undoped silicon standing between them — because the oxide mask, not the furnace, decided where the dopant could go.

Blanket Doping Versus Patterned Doping same furnace, same dopant, the entire difference is the mask 1954 / 1956: BLANKET DIFFUSION doped across the entire surface geometry defined later, by subtraction 1957: PATTERNED DIFFUSION three doped pockets, undoped silicon between them the dopant, the furnace, and the temperature schedule are unchanged; only the permission is new

## 3. Why Every Earlier Diffusion in This Project Could Not Have Been Selective

Compare this step directly to Steps 3 through 7 of the 1956 drift-transistor process, which used the same furnaces, the same dopants, and governed by the same physics as this step. The entire difference is the oxide pattern now sitting on the wafer when the furnace door closes. It is worth stating plainly what this step did *not* do: it did not make junctions shallower, more abrupt, or better controlled in depth than 1956 achieved. It changed only *where* diffusion is permitted to happen — and that single change, not any improvement in junction quality, is what makes it possible to build more than one device from a single furnace cycle. The economic consequence of that single change will turn out to matter more than any refinement to the diffusion physics itself ever could.

Step 13 does not diffuse anything differently than the 1956 process already knew how to diffuse; it diffuses it somewhere selective for the first time, and that is the whole payoff of the twelve steps that came before it.

Take photolithography oxide masking 1957 diffuse through windows 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