3D silicon crystal structure; each individual 
atom is bonded to four others. 
Credit: NSO

Semiconductors are an important part of technology and especially imaging to a high quality - CCDs (Charge-coupled Devices) are semiconductor devices. Unlike conductors such as metals, semiconductors are not able to naturally conduct electricity due to the movement of the free electrons within the material. The most common material used in semiconductors is silicon; a material that is not a natural conductor of electricity unless certain conditions are met.

If silicon atoms are stacked to form a silicon crystal with a regular structure, all of the silicon atoms form covalent bonds with one another - this holds the crystal together (covalent bond: the sharing of an electron pair between atoms). To transform the silicon into a material that can conduct electricity, free electrons must be available to move within the crystal. Silicon only needs a low-energy input to move a number of electrons from the valence band to the conduction band to meet the conditions to conduct electricity.


The conductivity of a semiconductor can be increased through a process known as doping. There are two different types of doping; both involve adding an impurity (a substance that is not silicon) to the silicon crystal. They are known as n-doping and p-doping:


n-doping: Silicon is doped with a pentavalent atom (five electrons in its outer shell) such as Phosphorus (P) which replaces a small amount of Silicon atoms in the crystal lattice structure. Each silicon atom before the crystal was doped shared a covalent bond with four neighbouring silicon atoms, the Phosphorus atom does the exact same thing – this leads to the fifth Phosphorus electron that can bond (but is unbound to a neighbour) being very loosely bound inside the crystal and can be freed very easily; making it a free electron to conduct electricity. The electrons that are freed have negative charge, thus the process is called n-doping.


2D representation of how Silicon (Si) atoms will bond to a Phosphorus (P) atom in a Silicon crystal structure. 
The P atom has a spare electron that is not bonded to anything and is therefore loosely bound to the P atom. 
Note: The rings only represent the electrons in the outer shells of the atoms.
Credit: NSO


p-doping: Silicon is doped with a trivalent (three electrons in its outer shell) impurity such as Boron (B). Like n-doping, the Silicon lattice structure wishes to bond four silicon atoms again to the Boron atom. Three electrons are available to bond from the Boron atom, but a fourth bond still wants to be made – this comes from a neighbouring atoms’ bound electron which creates a hole in the parent atom's outer shell as the electron moves over. Once this hole is freed up, another neighbouring electron will also wish to takes that one's place….thus the hole moves throughout the structure!


1. The three electrons in the outer shell of the B atom form covalent bonds with
three of the surrounding Si atoms, but there is one bond unfilled.
Credit: NSO


2. An electron from another atom/bond jumps to fill the hole, but once this hole
is filled another immediately takes its place; the hole moves through the lattice. 
Credit: NSO


This however does not quite give us a conductive material just yet..


Joining n-doped and p-doped material together

When an n-doped and p-doped material are joined, we have something called a diode. The contact of the two doped materials creates a junction where the free electrons in the n-doped material will diffuse across to fill the holes in the p-doped material, and oppositely, the holes in the p-doped material will diffuse across to recombine with the free electrons! This process switches the charge of the n-doped region to positive (has lost its free negatively charged electrons) and the p-doped region to negative (has gained negatively charged electrons). An electric field is then created due to the charge on either side of the junction; the small area around the junction is called a depletion region: this stops any further electrons from diffusing across to the p-doped region unless an applied voltage (external energy) helps the extra electrons overcome the coulomb barrier of the space charge in the depletion region.


n-doped and p-doped material being joined; the electrons move to fill the holes in the p-doped region and the holes
move to recombine with the electrons in the n-doped region.
This creates a depletion region for a distance on either side of the junction (the joining) which then builds a coulomb
barrier, stopping additional electrons from passing through unless they have more energy
Credit: NSO


How does this work in a CCD chip?

CCD chips are constructed from an assortment of doped silicon materials, but the summary of the process that converts photons to charge can be described simply:

Two words: the Photoelectric Effect.  When an incident photon hits the loosely bound electrons in the n-doped material, it gains energy and is referred to as a photoelectron. The extra energy is enough for an electron to hop out of its loosely bound state and become a free electron. If there are a lot of photons then there are a lot of loosely bound electrons being freed with enough energy to overcome the aforementioned coulomb barrier and be measured as an electrical current.

The depletion region is also a place where charge can be stored. There Is however a limit to how much charge can be stored in a depletion region/how many photoelectrons can recombine with holes which is referred to as a potential well. A deeper potential well means more charge can be stored and pixels will not saturate as easily as they can collect more charge before 'overfilling' – to do this one would simply dope the two different silicon-based materials more and more to increase the amount of possible electron-hole pairs and thus increase the depth of the potential well. This can’t however be done indefinitely; excessive doping can increase the dark current (thermal process in the chip due to the generation of new electrons) because of the electric field created at the junction – this will add noise to the resulting data.


For more in-depth reading on the physics of valence bands, conduction bands or any physics terms that are slightly unfamiliar and their relation to semiconductors and CCDs that we've mentioned – we’d recommend this document by the University of Manitoba:

Operation of CCD chip: