Ion Implantation and Doping

Every transistor needs precisely placed regions of N-type and P-type silicon — that’s the whole basis of how a MOSFET turns on and off. But how do manufacturers actually get dopant atoms into exactly the right spots, at exactly the right depth, without disturbing anything else on a wafer covered in delicate patterned structures? The answer is ion implantation: firing dopant atoms at the wafer like a particle accelerator, one atom at a time.

From “doping” the concept to doping the process

The foundations section introduces doping as an idea: add a tiny number of phosphorus or boron atoms into silicon’s lattice, and you create extra free electrons (N-type) or extra “holes” (P-type). That page treats doping mostly as a conceptual substitution — swap in an atom with one extra or one fewer valence electron.

Manufacturing doping is that same idea, but with an enormous added constraint: it has to happen in exactly the right place, at exactly the right depth, in exactly the right concentration, on a wafer that already has other structures built on it (like a transistor’s gate), without melting or damaging anything nearby. Older techniques like simply diffusing dopant gas across a heated wafer struggle to hit that level of precision. Ion implantation was developed specifically to solve it.

🧩 Think of it like… the difference between mixing sugar evenly into a whole bowl of batter (diffusion doping) versus using a precision airbrush to inject exact, tiny amounts of sugar at exact depths into specific, marked spots on a cake that's already decorated (ion implantation). The airbrush approach is slower and needs expensive equipment, but it's the only way to hit a target that precise without ruining the decoration around it.

How ion implantation works

Ionize — a source gas containing the dopant element (such as boron trifluoride for boron, or phosphine for phosphorus) is ionized into charged atoms (ions).
Select — a magnetic field bends the stream of ions, and only ions with the correct mass-to-charge ratio (i.e., the right dopant species) are allowed through — this is mass separation, and it keeps out unwanted contaminant ions.
Accelerate — the selected ions are accelerated through a strong electric field, reaching energies from a few thousand to a few million electron-volts, giving them enormous speed.
Implant — the high-speed ion beam is scanned across the wafer. Wherever the wafer surface is exposed (through openings in a patterned photoresist or hard mask left over from photolithography), ions physically slam into the silicon and come to rest just beneath the surface. Areas still covered by resist or mask block the ions from reaching the silicon underneath.
Anneal — the wafer is heated briefly (rapid thermal annealing) to repair the crystal lattice damage caused by the ion impacts and to move the implanted atoms into proper substitutional lattice sites, where they can actually contribute a free electron or hole.

    flowchart LR
    A["Dopant source gas"] --> B["Ionize into charged atoms"]
    B --> C["Mass-select correct ion species"]
    C --> D["Accelerate to high energy"]
    D --> E["Implant through mask openings"]
    E --> F["Anneal: repair lattice,<br/>activate dopants"]

Cross-section: implanting through a mask

high-energy dopant ion beam

maskmasksilicon waferimplanted dopant region

Only the ions traveling through the open gap in the mask reach the silicon; the rest are absorbed by the resist or hard mask before they ever get close. The depth the ions penetrate is controlled by their energy — higher energy drives them deeper — and the concentration is controlled by the dose, the total number of ions delivered per unit area.

Why implantation, not just diffusion

Older chips used simple thermal diffusion: expose the wafer to a hot dopant-containing gas or solid source and let dopant atoms slowly migrate into the silicon by heat alone. Diffusion is cheap but hard to control precisely — depth and concentration are coupled together, controlled only indirectly by temperature and time, and the doping profile spreads out gradually rather than sitting in a sharply defined region. Ion implantation decouples these variables: energy sets depth, dose sets concentration, and the mask sets location, independently and repeatably. That control is essential for modern transistors, whose active regions can be only tens of nanometers across.

Q = \frac{I \cdot t}{q \cdot A}

Here $Q$ is the implant dose (ions per square centimeter), $I$ is the beam current (how many ions per second are delivered), $t$ is exposure time, $q$ is the electric charge on each ion, and $A$ is the implanted area. This is simply “how much charge hit the wafer, converted into how many ions per unit area” — and it’s the dose that ultimately sets how heavily doped, and therefore how conductive, the implanted region becomes.

The same ion implanters are also used to create very shallow, heavily doped regions right at a transistor’s source and drain edges (sometimes called “halo” or “extension” implants), which are critical for controlling the razor-thin channel of modern transistors — see The MOSFET for why that channel region matters so much.

Key takeaways

Ion implantation dopes silicon by physically firing charged dopant atoms into the wafer at high energy, unlike simple heat-driven diffusion.
The process is ionize → mass-select → accelerate → implant through a mask → anneal to repair crystal damage and activate the dopants.
Implant energy controls depth, and dose (ions per unit area) controls concentration, giving independent, precise control unavailable with plain diffusion.
This precision is what allows modern transistors, with active regions only tens of nanometers wide, to have exactly the doping profile the design requires.

Etching and Deposition Metallization and Interconnects