what is ion implantation

**Ion implantation is how a fab actually changes silicon's electrical personality, not just its shape.** Etch and CVD sculpt and build up physical structure, but neither one turns plain silicon into something that can switch current on and off. That job belongs to doping: firing specific dopant atoms — boron, phosphorus, arsenic — into the wafer at high speed so they lodge inside the crystal lattice and donate or accept electrons, creating the p-type and n-type regions that become a transistor's source, drain, and channel. **The implanter itself is essentially a small, precise particle accelerator aimed at a wafer.** A solid or gaseous dopant source is ionized into charged particles, then a mass-analyzing magnet bends the ion beam and filters out everything except the exact dopant species wanted — this is what keeps a boron implant from being contaminated with some other stray ion. The selected ions are accelerated to a chosen energy (higher energy drives them deeper into the wafer) and scanned across the surface until the target dose — ions per square centimeter — is reached. Energy sets *how deep* the dopant lands; dose sets *how much* dopant is there. ```svg Ion Implantation: Beamline to Depth Profile A schematic of an ion implanter beamline from ion source through mass-analyzing magnet to the wafer target, alongside a depth-concentration profile of the implanted dopant before and after anneal. ION IMPLANTATION: BEAMLINE AND DEPTH PROFILE IMPLANTER BEAMLINE Ion Source Ionize dopant gas/solid Mass-Analyzing Magnet Selects exact dopant species Acceleration Column Energy sets implant depth Scanned Beam Dose = ions / cm² Wafer target tilted 7° to avoid channeling Energy controls depth; dose controls concentration — set independently Wafer tilt prevents ions from channeling straight down crystal planes uncontrolled DOPANT CONCENTRATION VS. DEPTH 0 depth → as-implanted (damaged lattice, narrow peak) after anneal (lattice repaired, dopant activated, slight spread) ``` **A wafer straight out of the implanter is technically doped but not yet usable — the damage has to be repaired first.** The implant process is violent at the atomic scale: each incoming ion knocks silicon atoms out of their crystal lattice positions as it slows down and stops. Left alone, that damaged, partially amorphous silicon has terrible electrical properties and most of the dopant atoms sit in the wrong lattice positions to actually donate or accept a charge carrier. A thermal anneal step — often a rapid spike anneal reaching over 1000°C for only a second or two — repairs the crystal structure and pushes dopant atoms onto proper lattice sites (activation), while a short, sharp temperature spike limits how far the dopants diffuse from where they were placed. | Dopant | Type | Typical Role | Diffusion Behavior | |---|---|---|---| | Boron (B) | p-type | Source/drain, well doping | Light atom, diffuses relatively fast — needs tight thermal budget | | Phosphorus (P) | n-type | Source/drain, well doping | Moderate diffusion, common general-purpose n-type dopant | | Arsenic (As) | n-type | Shallow source/drain junctions | Heavy atom, diffuses slowly — good for ultra-shallow junctions | | Antimony (Sb) | n-type | Ultra-shallow, low-diffusion junctions | Very heavy, minimal diffusion, used for tightest junction depths | ```flowchart st=>start: Select dopant species and generate ions at the ion source select=>operation: Mass-analyzing magnet filters beam to the exact desired ion species accelerate=>operation: Accelerate ions to target energy, setting implant depth scan=>operation: Scan beam across tilted wafer until target dose (ions/cm²) is reached anneal=>operation: Rapid thermal (spike) anneal repairs lattice damage and activates dopants verify=>operation: Metrology confirms dopant depth profile and sheet resistance pass=>end: Doped region meets target junction depth and electrical activation st->select->accelerate->scan->anneal->verify->pass ``` **As transistors shrink, the room for error in doping shrinks even faster.** A modern GAA nanosheet or FinFET needs an ultra-shallow, precisely activated source/drain junction — too deep and the device leaks or shorts to neighboring structures; too little activation and the transistor simply can't drive enough current. That's why implant energy, dose, tilt angle, and anneal profile are treated as tightly controlled, independently tunable knobs rather than a single "doping" step: at advanced nodes, getting any one of them slightly wrong is often the difference between a transistor that switches cleanly and one that doesn't work at all.

Go deeper with CFSGPT

Get AI-powered deep-dives, save terms, and run advanced simulations — free account.

Create Free Account