Cp-4 Chemical Polish
# CP-4 Chemical Polish: Diffusion-Limited Etching, Surface-State Pinning & the Accidental Inversion Layer
CP-4 was not chosen because it etched germanium — almost any acid mixture does that. It was chosen because it etched germanium smoothly, in a regime where the chemistry itself, not the operator's hand, controlled the result. A diamond saw and a lapping slurry leave a germanium surface mechanically torn: a band of micro-cracks, dislocations, and dangling bonds several micrometers deep that would otherwise quench the minority-carrier lifetime the point-contact transistor's entire amplification mechanism depends on. CP-4 — a mixture of nitric acid, hydrofluoric acid, glacial acetic acid, and a trace of bromine — removed that damaged layer and, in doing so, left behind something nobody had deliberately engineered: a naturally occurring p-type inversion skin on the n-type bulk, created entirely by the density of electronic states the freshly etched surface itself presented. Understanding why CP-4 produces a mirror-smooth surface, and why that surface spontaneously inverts, explains a step every subsequent point-contact process flow decision — whisker spacing, forming voltage, encapsulation — was built on top of.
---
## 1. Why Three Reagents and Not Just One Acid
A single strong acid etching germanium tends to attack whatever surface irregularity reacts fastest — exposed dislocations, scratch bottoms, grain edges — which *amplifies* roughness rather than removing it, since the fastest-etching spots are consumed fastest and stay recessed. CP-4's formulation splits the chemistry into separable roles specifically to avoid this: nitric acid handles oxidation only, converting surface germanium to germanium dioxide; hydrofluoric acid handles dissolution only, converting that oxide into water-soluble fluogermanic acid; and glacial acetic acid dilutes and moderates the overall reaction so that neither step runs away locally. Bromine is present only in trace quantity and serves as a rate catalyst, accelerating and — critically — *evening out* the oxidation step across the surface rather than letting it localize at defect sites:
Splitting one net etch reaction into an oxidation half-step and a dissolution half-step — each handled by a different reagent — is what converts "acid attacks germanium" into a controllable, reproducible polish rather than an uncontrolled, roughness-amplifying dig.
---
## 2. Diffusion-Limited Kinetics: Why CP-4 Polishes Instead of Roughening
Whether an etch smooths or roughens a surface depends on which step is rate-limiting. If the surface chemical reaction itself is the bottleneck (reaction-limited etching), the etch rate is highly sensitive to local crystal orientation and defect density, and a rough surface tends to stay rough or get rougher — fast-reacting high-energy sites etch faster than their surroundings. If, instead, the rate-limiting step is how quickly fresh reagent can diffuse through a thin stagnant boundary layer at the surface to reach the reaction site (diffusion-limited etching), the etch rate becomes nearly independent of local crystal orientation or defect state, because every point on the surface is waiting on the same comparatively slow transport process rather than competing reaction kinetics. A protruding high point, sitting slightly further into the bath's bulk fluid, actually receives fresh reagent *faster* than a recessed low point shielded by the boundary layer — so diffusion-limited etching preferentially attacks high spots and smooths the surface as it proceeds:
where $D$ is the reagent's diffusion coefficient, $\delta$ is the boundary-layer thickness, and $C_{\text{surface}} \approx 0$ when the surface reaction is fast relative to transport. CP-4's acetic-acid dilution and the HF-dissolution step were tuned specifically to push the overall process into this diffusion-limited regime — which is the entire reason it is described as a *polish* rather than merely an *etch*, and why mild agitation of the bath (thinning $\delta$ uniformly) was part of the standard process rather than a variable left to chance.
---
## 3. Fermi-Level Pinning and the Accidental p-Type Skin
Bulk n-type germanium has its Fermi level close to the conduction band, reflecting an abundance of free electrons from antimony donors. But a freshly CP-4-etched surface carries an extremely high density of dangling-bond and adsorbed-species electronic states within the bandgap — $D_{it} > 10^{13}\,\text{cm}^{-2}\text{eV}^{-1}$, vastly more states than the bulk doping concentration could ever fill or empty by comparison. A state density that large pins the surface Fermi level at whatever energy the states themselves sit at, regardless of what the bulk's doping would otherwise dictate — and for germanium's native surface states, that pinning energy sits close to the valence band. The bands must bend upward approaching the surface to connect this pinned surface condition to the flat, electron-rich bulk a few hundred nanometers down, and that bending pushes the surface hole concentration above the surface electron concentration: the n-type crystal's own surface becomes p-type, not because anyone doped it that way, but because the state density the etch itself exposed was too large for the bulk's donors to override.
This inversion layer is precisely what John Bardeen's surface-states theory identified as the reason early field-effect experiments at Bell Labs had failed before the point-contact transistor succeeded — and it is also the layer a reverse-biased collector whisker, pressed onto this same surface, actually makes contact with.
---
## CP-4's Place in the Point-Contact Process Flow
| What CP-4 Removes/Creates | Why It Matters Downstream |
|---|---|
| Saw/lap damage layer | Restores minority-carrier lifetime the whisker-collection mechanism needs |
| Surface roughness | Diffusion-limited polish gives whiskers a reproducible, uniform contact plane |
| p-type inversion skin | Supplies the actual electronic surface a collector whisker reverse-biases against |
| Native oxide (removed, regrows thin) | A thin re-oxide after rinse is tolerated; CP-4 does not leave the thick damage-era oxide behind |
Crystal growth and slicing hands CP-4 a geometrically correct but still-damaged die; CP-4 hands the next step — base stud soldering and whisker placement — a surface that is not only clean and smooth, but electronically primed with exactly the kind of naturally inverted p-type layer the point-contact mechanism was, in the end, built around.
Read CP-4 through a *kinetics-regime-not-recipe* lens rather than a *cleaning-step* lens: the specific reagent split and dilution were chosen to force the etch into a diffusion-limited regime where polishing, not pitting, is the geometric outcome — and the inversion layer that same etch leaves behind, an accident of surface-state density rather than a deliberately introduced dopant, turned out to be the actual electronic surface the entire point-contact amplification mechanism depended on.