First Pellet Drop

# First Pellet Drop (p): Overwhelming an N-Type Melt on Purpose

The first junction in the grown-junction sandwich is not created by adding anything to the crystal — it is created by hijacking the melt the crystal is still drinking from. Until this instant, the Czochralski puller has been drawing a continuous n-type germanium crystal out of an antimony-doped melt, exactly as it was doing from the moment of seed dip. The first pellet drop changes nothing about the puller, the pull rate, or the crystal already grown — it changes only the melt's chemistry, by adding a measured slug of gallium that outnumbers the antimony already dissolved there by a wide margin. Every atom of germanium that solidifies after this moment freezes out of a melt that has flipped from n-type to p-type, and the crystal — which cannot tell the difference between "old melt" and "new melt," only between "what it's touching right now" — simply keeps growing, now building p-type material directly on top of the n-type material it already laid down. The junction is the handoff line between those two moments, and it exists without a mask, a diffusion furnace, or a human hand anywhere near the crystal itself.

## 1. Why "Overwhelms," Not "Converts" — the Doping-Ratio Threshold

$$N_A^{\text{melt}} \gg N_D^{\text{melt}}$$

The melt does not become pure gallium-doped; the antimony already dissolved in it does not disappear. What has to happen is simpler and cruder: the net acceptor concentration contributed by the dropped gallium has to exceed the existing donor concentration by enough margin that the melt's *net* carrier type flips cleanly from n to p, with no ambiguous transition region wide enough to show up as a measurable defect in the finished device. Because the melt is well-mixed by convection and crucible rotation, this is a bulk concentration problem, not a local one — the pellet has to fully dissolve and disperse before the net doping sign at the growing interface is reliably p-type everywhere across the crystal's diameter, not just near wherever the pellet happened to land.

$$N_A^{\text{melt,eff}} - N_D^{\text{melt,eff}} = \frac{m_{\text{Ga}}}{M_{\text{Ga}}} \cdot \frac{N_{\text{Av}}}{V_{\text{melt}}} - N_D^{\text{(Sb, pre-existing)}}$$

Pellet mass $m_{\text{Ga}}$ is sized against the known melt volume $V_{\text{melt}}$ and its pre-existing antimony concentration precisely so that this difference lands comfortably positive — not marginally positive, because a melt sitting close to the n/p threshold is exactly the condition that produces a soft, graded junction instead of the abrupt one the device needs.

## 2. Real Diagram: What the Pellet Drop Actually Does to the Melt

First Pellet Drop — Flipping the Melt While the Pull Continues Ga pellet dropped into the still-pulling melt — the crystal keeps growing, now freezing from p-type liquid before the drop n-type melt Sb carries the charge n crystal pulling at v_p crystal and melt both still n-type, pull unchanged after the drop p-type melt Ga now dominates Ga pellet, dissolving n crystal pull continues, unchanged new growth freezes p-type — junction forms at this instant pellet dropped here

## 3. The Junction Is a Timestamp, Not a Boundary Someone Draws

Nothing about the crystal's lattice, its diameter, or its growth rate changes at the handoff instant — the only thing that changes is which melt the next atomic layer freezes out of. This is the structural difference between a grown junction and every junction formed afterward in semiconductor history: an alloy-junction or diffused-junction device gets its junction depth from a *separate, measurable process step* (furnace time, diffusion coefficient, regrowth distance) that happens to an already-solid wafer. A grown junction gets its depth from nothing but *elapsed pull time after the pellet drop*, converted to length through the pull rate already fixed at seed dip:

$$x_j = v_p \,(t - t_{\text{drop}})$$

There is no independent knob for junction depth here separate from the pull-rate schedule set in step one — which is exactly why the gallium pellet's mass, and the melt volume it's dropped into, have to be chosen in advance with enough margin that the flip happens close enough to instantaneously that $x_j$ is a well-defined number rather than a diffuse transition zone smeared across a measurable fraction of the pull.

Net Carrier Type Along the Ingot, Across the First Drop x_j = v_p (t − t_drop) — a sharp flip means a well-defined junction depth pull length along ingot → net carrier type (n above, p below) n-type, N_D (Sb) p-type, N_A (Ga) pellet drop, t_drop x_j, if transition is sharp soft flip if margin too thin — smears x_j

## First Pellet Drop's Place in the Process Lineage

First pellet drop is step two of the six-step grown-junction transistor device process flow — immediately after seed dip & pull start, and the first of two deliberate melt-dosing events (this one gallium, the next antimony) that carve the n-p-n sandwich into a single continuously-grown crystal, ahead of ingot completion, crosswise slicing, and lead attachment.

Take first pellet drop 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