Doping

Doping is powerful because it directly sets the number of mobile carriers, and carrier count controls conductivity. A few relationships make this precise.

Two methods put dopants into silicon. They differ fundamentally in how the dopant gets in and how precisely its amount and depth can be controlled.

• Ion implantation: dopant atoms are ionised and fired into the wafer as an accelerated beam. The amount (dose) and depth (energy) are set electrically and independently, giving precise, repeatable control — even at low temperature.
• Diffusion: the wafer is heated and dopants move into the silicon from a source at the surface, driven by concentration gradient and temperature. Amount and depth are coupled through temperature and time, so they are harder to control independently.
• The fundamental difference: implantation delivers a measured quantity to a chosen depth by controlling a beam; diffusion relies on thermally-driven movement from the surface, where deeper almost always means more spread-out.

Ion implantation is the workhorse for controlled doping. Conceptually it is a beam of dopant ions aimed at the wafer.

1. Ion generation→
2. Acceleration→
3. Beam (ion selection & scan)→
4. Wafer→
5. Implanted region

• Ion generation: the dopant element is turned into ions (charged atoms).
• Acceleration: an electric field accelerates the ions to a chosen energy.
• Beam: the desired ion is selected and the beam is scanned across the wafer so the dose lands uniformly, only where the mask allows.
• Dose: the number of ions per unit area — the main control over how heavily the region is doped.
• Energy: largely sets how deep the ions come to rest, and thus the junction depth.
• Depth & profile: together, energy and dose shape how dopant concentration varies with depth — the region's profile.

Diffusion introduces dopants by letting them migrate into hot silicon — the same way a drop of dye spreads through warm water, but atom by atom into a crystal.

• Thermal diffusion: at high temperature, dopant atoms at the surface move into the silicon, seeking lower-concentration regions.
• Concentration gradients: dopants flow from where they are plentiful (the surface source) toward where they are scarce (deeper in), so the gradient drives the process.
• Temperature dependence: diffusion is strongly temperature-sensitive — hotter means faster movement and deeper penetration.
• Diffusion profiles: the result is typically a graded profile, highest at the surface and tapering with depth.

Implanting ions damages the crystal and leaves many dopant atoms sitting in the wrong places. A thermal anneal fixes both — but heat is a double-edged tool.

• Lattice damage: incoming ions knock silicon atoms out of place, disordering the crystal near the surface; heavy damage degrades how carriers move.
• Dopant activation: heating lets dopant atoms settle onto proper lattice sites, where they finally contribute free carriers — an un-activated dopant does nothing electrically.
• Diffusion during annealing: the same heat that activates dopants also lets them move, so the doped region can spread and the junction shift during the anneal.
• Trade-offs: enough heat is needed to repair damage and activate dopants, but too much thermal budget spreads dopants and blurs shallow junctions — so modern processes favour fast, brief anneals (e.g. rapid thermal annealing) to activate while limiting diffusion.

Put a p-type region next to an n-type region and something remarkable happens: you get a PN junction, the building block of diodes and the heart of every transistor. This is doping's big payoff.

• Where p meets n, free electrons from the n-side and holes from the p-side meet near the boundary and cancel, leaving a thin zone with almost no mobile carriers — the depletion region.
• The fixed dopant ions left behind in that zone (positive on the n-side, negative on the p-side) create a built-in potential — an internal voltage step across the junction, even with no battery attached.
• Carrier movement: that built-in field opposes further crossing, setting up a balance. Applying an external voltage one way shrinks the barrier and lets current flow; the other way widens it and blocks current.
• Diode behaviour: this is exactly why a PN junction conducts in one direction and blocks the other — a diode. Combine junctions and you can build a transistor that switches and amplifies.

A transistor is essentially an arrangement of doped regions. Doping type, amount, and placement define every part of it.

• Source and drain: heavily-doped regions of one carrier type that the current flows between.
• Channel: the region between source and drain whose conduction the gate switches on and off; its doping helps set the transistor's threshold.
• Wells: larger regions of a given doping type that host transistors of the opposite type, isolating them and providing the body the device needs.
• Junctions everywhere: source-to-well and drain-to-well are PN junctions, so the whole device is built from the doped regions and the junctions between them.

A doped region is not a uniform block — dopant concentration varies with depth and position, and the shape of that variation is what engineers actually design.

• Concentration profile: how dopant concentration changes with depth. The peak location and how quickly it falls off define the region's electrical behaviour, not just the total amount added.
• Shallow vs deep junctions: a shallow junction keeps the doped region close to the surface (increasingly required for small transistors to keep the gate in control), while a deep junction extends further. Shallow junctions are harder to make and tend to have higher resistance — a trade-off.
• Abrupt vs graded junctions: an abrupt junction changes from one doping type to the other over a very short distance; a graded junction changes gradually. This shape affects the junction's electric field, its capacitance, and how it breaks down.
• Dose vs concentration: dose is the total dopant delivered per unit area (the whole profile added up); concentration is the local density at a given depth. The same dose can give very different peak concentrations depending on how spread out the profile is.
• Lateral vs vertical spread: dopants spread not only downward (vertical) but sideways under the mask edge (lateral), especially during annealing. Lateral spread sets the effective spacing between regions and matters more and more as devices shrink.

What decides how deep implanted ions go, and how spread out they end up? A few physical ideas explain the profile — described conceptually, not as equipment settings.

• Ion energy: the kinetic energy given to the ions. Higher energy generally means the ions penetrate deeper before stopping.
• Stopping mechanisms: ions lose energy two ways — by interacting with the target's electrons (electronic stopping, dominant at high energy) and by colliding with atomic nuclei (nuclear stopping, dominant at low energy and responsible for most lattice damage). An ion stops when its energy is used up.
• Implantation depth: the average depth at which ions come to rest (the projected range), set mainly by the ion, the target, and the energy.
• Straggle: ions do not all stop at exactly the same depth — there is a statistical spread around the average. This straggle gives the profile its width, both in depth and laterally.
• Channeling: if the ion beam lines up with the crystal's open rows, some ions slip down these 'channels' and travel far deeper than expected, distorting the profile. It is managed by tilting the wafer or disordering the surface first so the channels are blocked.

Implantation delivers dopants but also damages the crystal and leaves atoms off their sites. Annealing fixes both — yet the same heat that fixes them also moves them.

1. Implantation damage→
2. Annealing (heat)→
3. Lattice recovery→
4. Dopant activation

• Implantation damage: incoming ions knock silicon atoms out of place, disordering the near-surface crystal (heavy doses can even make it amorphous).
• Annealing: heating gives atoms the energy to move back toward their proper positions.
• Lattice recovery: the crystal re-orders (recrystallizes), removing damage that would otherwise trap or scatter carriers.
• Dopant activation: dopant atoms settle onto substitutional lattice sites, where they finally donate or accept carriers — an un-activated dopant contributes nothing.

A MOSFET is a voltage-controlled switch: a gate voltage decides whether current flows between two terminals (source and drain) through a channel. Doping defines every one of those parts — and the device's key properties.

• Source/drain engineering: the doping of the source and drain (concentration, depth, profile, and the lightly-doped extensions reaching toward the channel) sets their resistance and helps keep the gate in control as devices shrink. Tuning these regions is one of the biggest levers device engineers have.
• Threshold voltage: the gate voltage at which the transistor turns on. Channel and well doping strongly set it — too little and the device leaks when it should be off, too much and it is hard to switch on.
• Leakage: when off, the transistor's junctions and channel must block current. Doping errors, wrong profiles, or leftover implant damage cause leakage that wastes power.
• Mobility: how easily carriers move through the channel. Heavy doping and residual damage add scattering that lowers mobility — and thus drive current and speed — so more doping is not automatically better.
• Resistance: source, drain, and contact resistance fall as doping and activation rise, which speeds the device up — pulling against the mobility and leakage concerns above.

As transistors shrank, flat (planar) devices began to leak because the gate lost control of the channel. Newer 3D structures wrap the gate around the channel — and that changes how doping is used.

• FinFET: the channel is a thin vertical 'fin' with the gate on three sides, giving much better control. Because the gate controls the channel electrostatically, the channel is often lightly doped or nearly undoped, shifting emphasis onto precise, conformal source/drain doping of 3D surfaces.
• Gate-all-around (GAA): the gate fully surrounds the channel for even tighter control, so the same low-channel-doping idea is pushed further while doping the surrounding regions becomes more demanding.
• Nanosheet devices: a common GAA form using stacked horizontal sheets of channel wrapped by the gate. Doping these tiny, enclosed sheets uniformly and conformally — reaching all surfaces evenly — is a major challenge.
• Advanced architectures: the overall trend is more gate control over ever-smaller channel volumes, so doping must become shallower, more precise, and more conformal, with the action moving from doping the channel toward engineering the source/drain and contacts.

Like the rest of the fab, doping is not a fixed recipe but a controlled loop: set the inputs, measure the result, and correct the next wafers.

1. Dose control→
2. Energy control→
3. Annealing→
4. Metrology→
5. Electrical testing→
6. Feedback → next wafer

• Dose and energy are set and monitored to hit the target concentration and depth for each region.
• Annealing is controlled in temperature and time to activate dopants without over-diffusing them.
• Metrology measures the physical result — sheet resistance, concentration profile, and junction depth.
• Electrical testing confirms the devices actually meet targets such as threshold voltage, leakage, and resistance.
• Feedback: deviations feed corrections back into dose, energy, and anneal settings for following wafers (advanced process control), keeping the process centred as tools drift.

The topics below separate what is mature in production from what is still emerging or being pushed to its limits.

• Ultra-shallow junctions: making source/drain junctions extremely shallow yet highly activated and low-resistance at once — a continuing challenge as devices shrink.
• Advanced activation: activating dopants to very high levels without letting them diffuse, pushing beyond conventional equilibrium limits.
• Millisecond and rapid thermal processes: anneals lasting milliseconds or less (rapid thermal, flash, and laser annealing) that activate dopants while minimizing diffusion. Rapid thermal annealing is mature; the fastest laser/flash approaches are being pushed further for the shallowest junctions.
• Alternative doping approaches: conformal, surface-based doping methods that can dope complex 3D structures uniformly, and other non-implantation routes — emerging, driven by GAA and nanosheet devices.
• Compound semiconductor doping: dopants behave differently in materials beyond silicon (III-V semiconductors, silicon carbide, gallium nitride), an active area especially for power and high-frequency devices.