Doping: No, not the illegal in sports kind

The beginner's post for doping.

The first thing in your mind was probably the infamous incident of Lance Armstrong being caught using performance enhancing drugs during the Tour-de-France. Using these sorts of drugs is completely banned in sports and to enforce this the World Anti-Doping Agency was founded. In the context of semiconductors; doping is one of the most useful processes invented and has been utterly crucial in shaping the digital age.

Let me ask you a question, what is a semiconductor? It is a material with its conductivity value in between conductors (like copper, aluminum, and other metals) and insulators (like glass, PVC, ceramic, etc.). By definition, it seems like a pretty useless material. It is an inferior conductor and an insulator. Yet, the world runs on this stuff. Why is this the case?

The main property of a semiconductor is that we can change its conductivity by either optical or thermal excitation (heating or shining a light), or by doping. It follows the same thought process of performance enhancing drugs on the human body, hence the similarity in names, since doping in sports occurred earlier than the discovery of doping in semiconductors.

To make semiconductors more conductive, we try to substitute atoms in the crystal structure (an ordered arrangement of atoms, like a well-organized cabinet) with an element possessing more electrons, or less electrons. For example, silicon has 4 valence electrons. These electrons bond with other atoms to form compounds or with other silicon atoms to form a silicon structure. Changing it with an element like Boron which has 3 electrons will lead to one electron which is not bound to anything due to the absence of an electron to form a bond with. This absence of electron is termed as a hole. On the flip side, if we dope it with Phosphorus which has 5 valence electrons, then one electron is extra. This electron is free to conduct due to an excess of electrons in the structure. Doping with trivalent (like Boron) is called p-type doping as holes are categorized as positive charges. Doping with pentavalent atoms (like Phosphorus) is called n-type doping due to electrons being negative.

Figure 1: Hole and free electron formation alongside a snippet of the periodic table of just the three groups most used in semiconductor fabrication.

Figure 2: Location of doping in process flow.

There are two main ways to dope: diffusion doping and ion implantation. Diffusion is the process of mass travelling from high concentration areas to low concentration areas. The same way air pressure works, air travels from regions of high pressure to low pressure. Gases are introduced in the chamber, hydrides like Arsine (AsH₃), Phosphine (PH₃), or Diborane (B₂H₆). These gases thermally decompose due to high temperature of substrate which is a silicon wafer. Temperatures are usually 900-1200℃. The heated substrate also expands, opening up vacant spaces for dopant atoms to occupy. The dopant atoms are in high concentration outside the substrate and move inside due to the concentration difference. A person with high energy never really sits still, here the difference between atomic behaviour and a restless person is near zero.

Figure 3: Diffusion doping schematic.

(Reference: SemiCorex)

Figure 4: Models of atomic diffusion mechanisms for a two-dimensional lattice, with a being the lattice constant: (a) Vacancy mechanism. (b) Interstitial mechanism. Interstitial atom is one that occupies a site in a crystal structure that is normally unoccupied by atoms in that structure.

(Reference: https://www.cityu.edu.hk/phy/appkchu/AP6120/8.PDF)

Dopants penetrate quite deep depending on temperature, time, and material used. In silicon, maximum penetration can go beyond 100 micrometers for power transistors. In complementary metal oxide semiconductor (CMOS) technology, it is kept to around 1 micrometer. A hard mask is required for targeted doping. A hard mask is silicon nitride (Si₃N₄) or silicon dioxide (SiO₂), basically does the same work as photoresist but, as the name suggests, does not get eroded in high temperatures. The wafer is deposited with the mask material, then is sent for lithography and then the unrequired mask area is etched off. This adds an extra process step. It shares similarity with the idea of using a hard stencil for flat surfaces when painting.

The problems associated with this method are mostly targeted for CMOS. The high temperature causes atoms to keep moving and shifting around. Dopants may penetrate or go to a place that was covered by the mask due to collisions inside the wafer. This low control over positioning is a disadvantage. Basically, it spreads sideways. Junctions are not sharp and when working in such small scales, even a small leakage of charge can cause a huge cascade.

To work around this problem, a different technique was developed called ion implantation. Where doping worked like adding seasoning to hot food, ion implantation works like a bullet embedding itself in a wall. Depending on the energy with which it was shot out, the bullet goes that deep in the wall. Same theory is used in ion implantation. Ions are accelerated to a high velocity using a very high voltage (1-200 kV) and are shot towards the substrate. The ions penetrate the substrate to the desired depth. This is more controllable in terms of concentration and depth. This is used for CMOS technology. High temperatures burn the photoresist off, like in diffusion. A low temperature process like this one allows a photoresist of 1-2 micrometers to act as a mask. No need for a hard mask.

Figure 5: Ion implantation schematic.

(Reference: MKS)

Figure 6: Ion beam hitting the wafer.

(Reference: SemiCorex)

Ion implantation carries its own problems like crystal damage on the wafer. A bullet shot into a wall does not just phase through to the intended depth. The bullet creates a hole by displacing the material in the wall. Dopants will strike silicon atoms leading to displacement and defect generation. This is still an easy problem to fix as you just anneal the wafer. Anneal is a fancy word for controlled heating and cooling to repair structure damage in this case. The atoms gain thermal energy and just vibrate back into their locations, most of them.

Figure 7: Annealing as a healing process.

(Reference: Ebrary)

Figure 8: MATLAB plot of doping profiles where blue is gaussian and red is localized peak profile. Ion implantation has good penetration control as there are not many dopants near the surface while diffusion spreads downwards and sideways.

A bullet going into a wall will also remove some material. This process of removing atoms from a target using accelerated ions is called sputtering. This occurs in a very low amount in ion implantation so it can still be neglected. Ion implantation can cause surface charging leading to gate oxide breakdown. The gate is used to control the transistor. Vacuum needs to be created to allow the least number of collisions the ion suffers through to reach the wafer. Equipment using such high voltages requires special engineering. Diffusion requires a simple furnace while ion implantation requires very expensive and heavy machinery.

Doping requires a lot of formulae and mathematics to be able to model doping dosages and profiles correctly. I do not wish to bore you with mathematics but if you feel interested in learning more about doping then please check YouTube as many people have explained this process very well using animations.

Go back to main page