Diffused Base 1954 Select Usable Crystal Sections

# Select Usable Crystal Sections: One Surface Has to Carry Both Junctions, So One Standard Has to Cover the Whole Section

This step chooses which lengths of the inspected rod actually become wafers, and because this process diffuses both junctions inward from the same flat surface rather than working one dopant into one face and a different dopant into the opposite face, there is no way to accept a section that is good on one side and marginal on the other. An alloy-junction device's two alloyed disks sit on opposite faces of the same slice, so a section with slightly uneven quality across its thickness might still be usable if each face's local quality happens to suit the disk landing there. This process has no equivalent flexibility — one surface, one combined predeposition-and-drive-in schedule running twice, has to see uniform quality across the entire area it will ever diffuse into.

## 1. Acceptance Has to Be Area-Uniform, Not Just Length-Uniform

$$\text{Accept}(z) \iff \rho(z) \in [\rho_{\min}, \rho_{\max}] \ \text{and}\ N_t(z) < N_{t,\max} \ \text{across the full wafer area}$$

A section is only usable at axial position $z$ if resistivity $\rho(z)$ and defect density $N_t(z)$ both clear this process's guard bands *across the entire cross-sectional area* that will become a wafer — not just along a single measured line down the rod's length. Because both future junctions share this one surface, a local cold spot or a resistivity gradient across the wafer's diameter produces a base width or a collector resistivity that varies across the finished device's own active area, a defect an alloy-junction process, with its spatially separated disks, would be less exposed to.

## 2. Real Diagram: A Radial Gradient Becomes a Device-to-Device Difference on the Same Wafer

One Wafer, One Surface, No Room for a Radial Gradient every device cut from this wafer inherits whatever resistivity sat under it at this step on-target drifted on-target drifted a wafer this step approves on a center-point measurement can still hide a radial gradient the later diffusion schedule will faithfully reproduce everywhere

## 3. The Guard Band Has to Anticipate Two Diffusion Schedules Reading the Same Baseline

$$N_{D,\text{final}}(x) = N_{D,\text{bulk}} + N_{\text{diffused}}(x)$$

Every later diffusion profile $N_{\text{diffused}}(x)$ gets added to whatever bulk doping $N_{D,\text{bulk}}$ this step actually accepted — which means an accepted section's resistivity is not just a static property this step checks off, it is the baseline every later concentration profile in Steps 12 through 19 will be measured against. A guard band set too loosely here does not just risk one bad collector; it risks every later junction-depth measurement being referenced against a baseline that was never actually uniform, making a perfectly executed diffusion schedule look inconsistent for a reason that traces back to this step alone.

Every Later Junction Depth Is Measured Against This Step's Baseline a shifted baseline looks like a diffusion-schedule error, even when the schedule itself ran perfectly depth into wafer, x → dopant concentration correct bulk baseline, this step's target actual baseline, missed by this step diffusion profile, identical schedule either way the crossing depth — the junction — ends up wrong even though the diffusion itself was executed exactly as planned

## Select Usable Crystal Sections's Place in the Process Lineage

Selecting usable crystal sections is step five of Bell Labs' twenty-six-step diffused-base manufacturing sequence — immediately after inspection confirmed where the rod's bulk quality actually sits, and before any section is cut into wafers. It is the step that commits a single, area-wide resistivity and defect-density baseline that every one of this process's later diffusion steps will silently assume held uniformly across the entire future device surface, a dependency this process carries more directly than an alloy-junction process, whose two disks never had to share one surface's baseline at all. Step six, slicing the germanium, works only with whatever sections this step actually approved.

Take diffused base 1954 select usable crystal sections 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