
Let´s talk about diffusion!
The diffusion is present in many semiconductor manufacturing steps. According to the semiconductor manufacturing step, the diffusion can be done with several materials and with different depths and surface concentrations.
Diffusion is a two-step process: deposition and drive-in. The first one was already discussed in my previous post. Today I will discuss only the drive-in.
As it is already known, diffusion is a mass transport phenomenon. The interfaces where it occurs can be solid, liquid, or even gaseous.
Diffusion, as a simplification of the process, occurs when a concentration gradient exists between two interfaces. There are a lot of great scientific explanations about the mechanisms that drive diffusions such as free energy, and chemical potential gradients, I will leave that out of this blog, because I will adopt, as the initial idea, a practical point of view, i.e., how it works in the industry.
For semiconductors, the main diffusion is the solid-state diffusion. In this case, the diffusion mechanisms are the diffusion in metals.
Inside the solids(metals), the diffusion occurs by atomic motion. To be able to do that we need two basic conditions: a free space adjacent to the existent atom and enough energy to break the chemical bonds responsible for the neighboring atoms union.
The two mechanisms used to explain atomic motion in solids are: substitutional and interstitial.
The substitutional is also known as a vacancy diffusion. In this case, the “new” atom moves to a vacant space, and at the same time, the vacancy moves to another space. This is the principal mechanism in the semiconductor.
The interstitial diffusion, the atoms migrate to adjacent interstitial positions not occupied by the lattice. It is faster than the vacancy diffusion and the impurity atomic ray is normally smaller than the crystalline structure atom.
These are the diffusion mechanisms present in semiconductor device production.
Fick´s second law drives the diffusion in solids. The time, temperature, and dopant concentration influence it – we can think about it as the first Fick´s law plus a continuity equation.
There are three types of crystalline structures used in the semiconductor industry: monocrystalline, polycrystalline, and amorphous. The wafer choice depends on which device will be produced, for example, power semiconductor, solar cell, MOS devices…
Now we can talk about the diffusion processes used to manufacture semiconductor devices. It is quite important to remember that diffusion is not a linear process, i.e., always occur a lateral diffusion during the process (not only in the x direction, but also in the y, and z directions).
The surface concentration of the doping species changes according to the semiconductor device to be produced. For power semiconductor devices we have higher dopant surface concentration and deeper diffusions than for MOS semiconductor devices.
The most used diffusion techniques are thermal and ion implantation.
Thermal diffusion is used when the lateral diffusion is not critical. For this diffusion type, there are three main processes: same diffusion species at both sides of the wafer done at the same run, both sides of the wafer are diffused at the same time but with different diffusion species, and only one species on one side of the wafer.
It´s important to remember that the choice of the thermal diffusion type will define the next process step, for example, if it will be needed or not to lap one side of the wafer to make another diffusion on this lapped side then it will be necessary to grow SiO2 as a protection on the other side of the wafer where you don´t want any contamination…
The ion implantation is a more precise process; it allows us to have a high level of control of the diffusion process (lateral, depth, and surface concentration). It doesn´t need high temperatures either, it´s a low-temperature process. Atoms are bombarded into the substrate, which produces defects in the lattice that demand an annealing after the implantation.
Ion implantation is a process used for shallow diffusion and low surface concentration and diffusion of small areas such as the transistor´s drain for example. For large areas such as power semiconductor devices, it turns out to be an expensive process.
Right now, that we already know the techniques used to do the diffusion, we can talk about the dopants as I mentioned before there are different materials to be diffused.
The different materials such as boron, phosphorous, gallium, aluminum arsenic, gold, and platinum are related to the process step we are doing at the moment. For the diode structure proposed in another post as a guide to discuss here, we can think about the main diffusion capable of creating the deep junction we need, if the device is fast recovery, we will need another diffusion to reduce the carrier’s lifetime: the diffusion to turn the device in a fast recovery rectifier.
For the main diffusion, we use phosphorous boron aluminum, and gallium. For the fast recovery time diffusion, it can be used platinum or gold and electron bombardment.
As always, the material´s choice will influence the next manufacturing process step.