Point Contact Transistor Grind Upper Lower Faces Flat
# Grind the Upper and Lower Faces Flat: One Device, Two Faces, Two Different Flatness Standards
Step 2 left behind an oversize block with a damaged shell on every sawn face and no flat reference surface anywhere on it. Step 3 grinds two of those faces — the one that will carry the point contacts and the one that will carry the base connection — down toward their finished dimensions. It would be easy to treat this as one operation applied twice, the same wheel and the same pass count on both sides for symmetry's sake. It is not. The two faces answer to different standards, because only one of them has to support two independent point contacts a few thousandths of an inch apart, and the other does not have to support anything nearly that demanding.
Flatness on the base face is a tolerance against a contact area, not against a distance. The base connection, fitted in Step 6, is a broad, low-resistance metal contact meant to spread current across the whole lower face rather than concentrate it at a point. A broad contact of that kind averages over local height variation the way a point contact cannot: a shallow grinding mark or a few microns of residual waviness simply becomes part of the metal-to-germanium interface area, contributing a negligible fraction of the total contact resistance. The base face therefore only needs to be flat enough that the mounting and metallizing operations downstream of it behave predictably — flat in the ordinary macroscopic sense, not flat relative to any feature on the device itself.
Flatness on the contact face is a tolerance against the point spacing, because that is the only distance the device actually has. The emitter and collector points that Step 5 will place on this face sit only a few thousandths of an inch apart, and the entire point-contact mechanism depends on an injected carrier crossing exactly that gap through the germanium immediately beneath both points. A surface irregularity the base face would never register — a pit, a ridge, a patch of uneven removal — can sit directly inside or beneath that gap, and because the device's active region is defined by the gap rather than by any larger reference dimension, the standard this face must meet scales with the gap, not with a fixed number carried over from some other process.
Every grind leaves behind damage this step did not ask for and the next step has to clean up. Material removal by an abrasive wheel does not stop neatly at the finished surface; it drives dislocations and microcracks some distance beneath it, and that depth scales with the grit size used — coarser abrasive removes stock faster but buries damage deeper, finer abrasive removes stock slower but confines the damage closer to the surface. This has nothing to do with the device's own geometry, and everything to do with what Step 4's etch allowance has to be sized against. Choosing too coarse a final pass on the contact face trades grinding time for an etch step that can no longer remove all the disturbed lattice before a point contact ever touches it — the same failure mode the machining allowance in Step 2 was meant to prevent, recreated one operation later by the grind itself.
The two faces can therefore specify different final grit even though they come off the same machine in the same setup. There is no requirement that both faces receive identical treatment, and treating them identically wastes capability in one direction while potentially under-delivering it in the other: a final pass fine enough for the contact face's point spacing, applied to the base face as well, spends grinding time the base contact will never notice; a final pass coarse enough to finish the base face quickly, applied to the contact face as well, risks leaving damage the etch allowance was not sized to remove. The process description's silence on this point is not an oversight — it leaves room for exactly this asymmetry, because the two faces are headed toward two different later steps with two different sensitivities.
| Step | Process operation | Input | Output | Specification | Constraint |
|---|---|---|---|---|---|
| 3.1 | Set the finished flatness and roughness standard for the base face | Base contact method planned for Step 6 | Base face grind specification | Flatness adequate for a broad, low-resistance metal contact over the whole face | Over-specifying this face wastes grinding time the base contact's averaging effect never collects on |
| 3.2 | Set the finished flatness and roughness standard for the contact face | Planned point spacing from Step 1/Step 5 | Contact face grind specification | Roughness small relative to the intended point spacing, not against a fixed general tolerance | A roughness standard borrowed from a wider-spaced device would under-specify this device's contact face |
| 3.3 | Choose grit progression and final pass for each face independently | Grind specifications from 3.1 and 3.2 | Grinding plan, per face | Final grit sized so its expected subsurface damage depth is within Step 4's planned etch allowance | A final grit too coarse for the contact face's spacing recreates the exact risk Step 2's machining allowance addressed |
| 3.4 | Grind the base face to its specification | As-cut body from Step 2, grinding plan from 3.3 | Base face ground flat | Measured flatness within the base face specification from 3.1 | A base face ground off-specification compromises contact uniformity before Step 6 ever applies metal to it |
| 3.5 | Grind the contact face to its specification | Base-face-ground body, grinding plan from 3.3 | Contact face ground flat | Measured roughness within the contact face specification from 3.2 | A contact face ground off-specification places point contacts on a surface whose local defects can sit inside the critical gap |
| 3.6 | Verify both faces meet their respective, independent standards before releasing to Step 4 | Ground body from 3.4 and 3.5 | Verified ground body | Each face checked against its own specification, not against the other face's | Checking only one face, or checking both against a single shared standard, can pass a body that fails the standard that actually matters for it |
Step 3 does not make either face finished; it makes each face answerable to the operation that is actually waiting for it. The base face only has to satisfy a contact method that averages over imperfection. The contact face has to satisfy a device geometry that cannot average over anything, because its entire mechanism lives inside a gap measured in thousandths of an inch. Grinding both faces to one shared standard would be the easier instruction to write and the wrong one to follow — it is only by letting each face answer to what comes next for it, rather than to the wheel that happens to be grinding it, that Step 4's etch allowance and Step 5's point placement inherit a surface actually built for them.