Point Contact Transistor Crystal Growth and Slicing
# Point-Contact Transistor Device Crystal Growth & Slicing: Zone Refining, Resistivity Targeting & Dice Preparation
Before a single whisker ever touched a germanium surface, the entire point-contact transistor process depended on a crystal-growth step that had nothing to do with transistors at all: purifying germanium to a level no prior industrial process had ever required. Bardeen and Brattain's device needed bulk resistivity controlled to within a narrow band and a surface free of the crystalline defects that would otherwise swamp the minority-carrier diffusion lengths their whole amplification mechanism depended on. That meant solving a purification problem first — repeatedly melting a zone of germanium and letting it refreeze so that impurities, which prefer to stay in the liquid rather than the solid, were swept along the ingot's length and concentrated at one end, discarded along with it — and only afterward dicing the purified crystal into the small, flat, defect-minimized dice that crystal growth and slicing actually hands off to the rest of the point-contact process flow.
---
## 1. Zone Induction Melting and Why Most Impurities Segregate Out
The germanium used in point-contact transistors began as germanium dioxide reduced in a hydrogen furnace into polycrystalline metal, then purified by passing a narrow molten zone repeatedly along the length of a bar held in a graphite boat inside a traveling radio-frequency induction coil. As each zone moves, it leaves recrystallized solid behind it and melts fresh material ahead of it — and because almost every electrically active impurity in germanium has a segregation coefficient well below one ($k_0 \ll 1$, meaning the impurity strongly prefers the liquid over the solid it is freezing from), each pass leaves the newly solidified region cleaner than the melt it came from, concentrating impurities into the zone of liquid still traveling ahead:
where $\ell$ is the zone length and $x$ is distance already swept by one pass. Running several zones in succession along the same bar — rather than relying on a single pass — compounds this effect multiplicatively, because each new pass starts from the already-improved profile the previous pass left behind rather than from the original unrefined material. The practical consequence for point-contact manufacturing: the usable portion of a zone-refined ingot was always a specific, retained fraction of the original bar, with the impurity-concentrated tail physically cut off and discarded before any wafer was diced from the material that remained.
---
## 2. Doping the Purified Melt to a Target Resistivity
Purification alone leaves germanium nearly intrinsic — useless for a device that needs a controlled, specific bulk resistivity. A separate, deliberate doping step followed zone refining: a measured quantity of a pentavalent donor, almost always antimony, was introduced into the melt during a final controlled-growth pass, targeting a donor concentration in the range $N_D \approx 10^{14}$–$10^{15}\,\text{cm}^{-3}$, corresponding to a bulk resistivity of roughly $1$–$10\,\Omega\cdot\text{cm}$:
This target band was not arbitrary. Too low a resistivity (too much donor doping) shortened the minority-carrier diffusion length $L_p$ that the point-contact device's entire current-gain mechanism — whisker spacing held comparable to $L_p$ — depended on; too high a resistivity made the bulk material resistive enough to add unwanted series resistance and slow the device's frequency response. Resistivity was checked directly on sample slices with a four-point probe before an ingot was approved to proceed to dicing, which meant the crystal-growth step effectively screened out material the rest of the point-contact process flow would otherwise have wasted whisker-placement labor on.
---
## 3. From Ingot to Flat, Defect-Minimized Dice
Why dice this small. A point-contact transistor's active region spans only the tens of micrometers between its two whiskers, so the die itself needed to be just large enough to hold a base contact and provide mechanical support — roughly $1.5\times1.5\times0.5\,\text{mm}$ was the practical minimum that could still be handled, soldered, and probed by 1947–1952 assembly technique. Larger dice wasted purified, doped crystal for no electrical benefit; smaller ones became too fragile to survive soldering to a base stud.
---
## Crystal Growth & Slicing's Place in the Wider Process Flow
| Step Family | What It Controls | Downstream Dependency |
|---|---|---|
| Zone refining | Bulk impurity concentration (purity) | Sets the ceiling on achievable minority-carrier diffusion length $L_p$ |
| Doping pass | Net donor concentration $N_D$, resistivity $\rho$ | Sets $L_p$ precisely within that ceiling — governs required whisker spacing $s$ |
| Diamond sawing | Wafer thickness, ingot yield | Introduces work damage the CP-4 etch must remove |
| Dicing | Die footprint | Sets mechanical handling limits for later whisker and stud assembly |
| Lapping | Final thickness, face flatness | Determines whether whisker tips make uniform, reproducible contact pressure |
Crystal growth and slicing is, in this sense, the step that decides — before any whisker, etch bath, or forming pulse ever touches the material — whether the rest of the point-contact process flow even has a chance of producing a working amplifier. A wafer with too high an impurity count, wrong resistivity, or excessive uncorrected saw damage cannot be rescued by careful whisker placement later; every subsequent step in the flow is working within the ceiling this one sets.
Read crystal growth and slicing through a *purity-and-geometry-budget* lens rather than a *raw-material-prep* lens: the zone-refining pass count, the doping target, and the saw/lap tolerances are not preliminary housekeeping before the real process begins — they are the first and most unforgiving control loop in the entire point-contact transistor device process flow, setting the diffusion length every later whisker-spacing and forming-pulse decision had to be made relative to.