Point Contact Transistor Chemically Etch Germanium
# Chemically Etch the Germanium: Removing the Damage the Grind Could Not Help but Leave
Step 3 ground the contact face to its own standard, but grinding cannot finish the job it starts — the abrasive that removes stock also drives a layer of disturbed lattice beneath whatever surface it leaves behind, and no amount of additional grinding removes that layer without simply creating a new one of its own. Step 4 is the operation that actually clears it, and it does so by a mechanism with no counterpart in the grinding step at all: dissolution that runs faster at a protruding peak than in a recessed valley, because a peak sits closer to fresh, unconsumed acid than a valley does. The surface does not just get thinner. It gets smoother, and it gets smoother through a process the previous step could not have substituted for no matter how fine its final abrasive pass.
This etch has only one face's damage budget to clear, because the base face was deliberately left untouched in Step 3. Step 3 ground the contact face to a fine standard sized against the eventual point spacing, and this step's etch time is sized to clear exactly that damage depth — not a second, independent budget for a second face, because there is no second ground face yet to account for. The base face will receive its own grind and its own etch later in the sequence, immediately before metallization, and that later operation will carry its own, separately sized etch allowance. Treating this step as though it had to answer for both faces at once would overstate its job; it answers for the one face Step 3 actually prepared.
What the etch exposes is not just a smoother surface, it is a chemically different one. A ground surface carries mechanically introduced lattice disorder and whatever contamination the grinding operation itself deposited — abrasive fragments, handling residue, oxide formed simply from exposure to air during and after grinding. The etch does not discriminate between removing disordered lattice and removing that surface contamination; both dissolve into the same bath. The surface this device meets going forward is therefore not the surface Step 3 left behind with the damage subtracted — it is bare germanium with no history on it at all, which matters here because this face has a long wait ahead of it before the point contacts arrive, and it needs to start that wait clean rather than carrying forward anything the grind left behind.
Over-etching is a real failure mode, not merely a wasted margin. Because the surface actively dissolves rather than merely abrading, every minute the block spends in the bath after the roughness target is reached removes material the device still needs — the body's finished thickness, set back in Step 2, shrinks further with every additional minute in solution. A process that etches for a fixed, generous duration regardless of how close the surface already is to its target trades a predictable amount of germanium thickness for a margin of smoothness the device does not require, repeating in a wet chemical step the same resource logic that governed the machining allowance on the saw.
| Step | Process operation | Input | Output | Specification | Constraint |
|---|---|---|---|---|---|
| 4.1 | Determine required etch depth from Step 3's grinding record for the contact face | Ground contact face from Step 3, grit and damage-depth estimate from 3.2 | Etch depth target | Etch depth at least the estimated subsurface damage depth for the contact face's final grinding pass | Underestimating the damage depth leaves disordered lattice the etch was supposed to remove |
| 4.2 | Prepare the etchant and establish a controlled etch rate | Etch depth target from 4.1 | Etch process parameters (concentration, temperature, agitation) | Etch rate characterized well enough that time can be used as a reliable control variable | An uncharacterized etch rate makes the stopping decision in 4.4 a guess rather than a measurement |
| 4.3 | Immerse the body and etch the contact face for the time required to clear the recorded damage depth | Ground contact face, etch process parameters from 4.2 | Partially etched contact face | Etch time sufficient to reach the target from 4.1 | Stopping too early leaves damage incompletely removed; the base face, not yet ground, is unaffected either way |
| 4.4 | Monitor and stop the etch once the roughness target is met on the contact face | Partially etched contact face from 4.3 | Etched contact face, at or past target | Measured roughness below the point-spacing-derived target from Step 3 | Continuing to etch past this point consumes finished-dimension germanium for no benefit this face collects |
| 4.5 | Rinse and dry to leave bare, uncontaminated germanium on the contact face | Etched contact face from 4.4 | Clean, bare contact face | No residual etchant, reaction byproduct, or handling contamination remaining on this face | A face left with residual contamination carries that history forward through the long wait before point contacts arrive |
| 4.6 | Re-verify finished body thickness against Step 2's target after this face's stock removal | Clean contact face from 4.5, target dimensions from Step 2.1 | Thickness-verified body | Finished thickness within the range Step 2 budgeted, accounting for this face's removal alone | The base face's own removal, still to come, has to be budgeted separately against the same Step 2 target |
Step 4 does not merely clean a ground surface; it finishes a job the grind was structurally incapable of finishing on its own, for the one face that is ready for it. Mechanical abrasion cannot remove its own subsurface damage without generating more of the same kind, which is why the sequence hands this operation to a chemical process with a genuinely different removal mechanism — one that happens to level the surface as a side effect of how it works, rather than as something it has to be separately instructed to do. The contact face leaves this step bare, clean, and ready for a wait the base face does not have to make, because the base face's own grind and etch are still ahead of it, timed to its own metallization rather than to this one.