Exhibiting at SEMICON Japan 2026 (Dec 9–11)? We are now accepting advertising inquiries from exhibitors.Ads for SEMICON Japan 2026Learn more

Comparison · Ion implantation and thermal processing

Ion implantation vs thermal diffusion: what is the difference?

Doping creates the n-type and p-type regions that make transistors and diodes work. Early silicon devices were doped entirely by diffusion furnaces. From the 1970s ion implantation took over most doping steps because it separates dose from depth and works at low temperature. Thermal processing did not disappear: every implant needs a subsequent anneal, and diffusion physics still determines the final dopant profile.

The short answer


Ion implantation ionises dopant atoms, accelerates them and fires them into the wafer, so the dose is set by the integrated beam current and the depth by the acceleration energy, independently and reproducibly. Thermal diffusion introduces dopants from a gas or solid source at high temperature and lets them spread by diffusion, so concentration and depth are linked through temperature and time and spread sideways under the mask. Implantation is the standard for modern CMOS; diffusion survives where deep, gradual junctions are wanted.

Side-by-side comparison


AspectIon implantationThermal diffusion
PrincipleDopant ions are accelerated and physically embedded in the crystalDopants enter from a surface source and spread by solid-state diffusion at high temperature
Dose controlMeasured directly by integrating beam current; precise and reproducibleDepends on source concentration, solubility, temperature and time
Depth controlSet by ion energy; very shallow to deep profiles possibleSet by diffusion length (temperature and time); shallow junctions are hard
Profile shapePeaked near the projected range; can be buried below the surfaceHighest at the surface and falling with depth
Lateral spreadSmall; masks define regions sharply (some lateral straggle)Large; dopant diffuses sideways under the mask edge
Crystal damageIon impacts displace atoms; annealing is requiredLittle, as the process is near thermal equilibrium
Process temperatureImplant near room temperature, then an activation annealHigh, typically around 800–1,200 °C for extended times
MaskingPhotoresist can serve as the maskNeeds a heat-resistant mask such as silicon dioxide
Typical usesWells, threshold adjustment, source/drain extensions and contacts, halo implantsDeep junctions in some power devices and solar cells, legacy processes

How each method works


An ion implanter ionises a source gas or vapour containing the dopant, such as boron, phosphorus or arsenic, selects the wanted ion with an analysing magnet, accelerates it and scans the beam across the wafer. Each ion slows down through collisions and comes to rest at a depth distributed around its projected range. Counting the charge delivered gives the dose directly.

Thermal diffusion traditionally has two stages. In predeposition the wafer is heated in a furnace with a dopant source, for example a gas such as phosphorus oxychloride or a doped glass, so that dopant enters the surface up to its solubility limit. In the drive-in step the source is removed and further heating pushes the dopant deeper. The resulting profile follows diffusion laws and is set by temperature and time.

Why CMOS moved to implantation


Scaled transistors need shallow source and drain junctions with steep edges and minimal sideways spread under the gate. Implantation controls depth through energy and dose through beam current, separately and precisely, and makes profiles possible that diffusion cannot form, such as buried peaks for wells or angled halo implants that suppress short-channel effects.

Diffusion, by contrast, spreads isotropically, so dopant moves under the mask edge by an amount comparable to the junction depth. In dense layouts that encroachment shifts transistor dimensions. Low process temperature is another advantage of implantation: photoresist can be used as a mask and earlier structures are not exposed to long high-temperature steps.

The anneal after implantation


Implanted ions displace silicon atoms and leave many dopant atoms in interstitial positions where they do not contribute carriers. An anneal repairs the lattice and moves dopants onto substitutional sites, a step called dopant activation. During the anneal, excess point defects from the implant damage can make dopants diffuse much faster than normal for a short time, an effect called transient enhanced diffusion.

To keep junctions shallow, advanced processes use very short anneals: rapid thermal annealing with spike profiles lasting about a second, and flash or laser annealing on millisecond or shorter timescales. Diffusion is therefore still part of every doping process; it is simply minimised rather than used as the main mechanism.

Where thermal diffusion is still used


Diffusion remains practical where junctions are deep and gradual, where high temperatures are already part of the flow, or where cost per wafer matters more than profile precision. Examples include some deep junctions and guard rings in power devices, emitters of crystalline silicon solar cells, and older device processes. In silicon carbide, by contrast, dopants barely diffuse even at high temperatures, so implantation is used for selective doping.

FAQ


What sets the dose in ion implantation?

The beam current integrated over time, divided by the implanted area and the ion charge. Because the charge is measured directly, typically with Faraday cups, dose can be controlled precisely and reproducibly. Acceleration energy sets depth, not dose.

Is thermal diffusion still used?

Yes, for deep junctions in some power devices, for solar cell emitters and in legacy processes. In advanced logic and memory, almost all intentional doping is done by implantation or by doping during epitaxial growth.

Why is annealing needed after implantation?

To repair lattice damage caused by the ions and to move dopant atoms onto lattice sites where they act as donors or acceptors. Without annealing, much of the implanted dopant would be electrically inactive and the damaged layer would have poor electrical properties.

Does implantation avoid diffusion entirely?

No. Dopants diffuse during every subsequent high-temperature step, and implant damage temporarily accelerates diffusion. Process engineers budget this thermal history and use very short anneals to keep profiles close to the as-implanted shape.

Terms compared


Japanese companies involved


More ion implantation and thermal processing comparisons


Sources


  1. Nissin Ion Equipment: Ion implantation dictionary, diffusion coefficient (Japanese)
  2. Nissin Ion Equipment: Ion implantation dictionary, projected range (Japanese)
  3. Nissin Ion Equipment: Ion implantation dictionary, enhanced diffusion (Japanese)
  4. Surface Finishing Society of Japan: Ion implantation, vol. 30 no. 5 (Japanese)(PDF)

More on this topic


Semiconductor comparisons

Ion implantation and thermal processing terms

Last updated: November 2, 2026