Doping: How Deliberate Impurity Makes Semiconductors Work
Perfectly pure silicon is a disappointing material. At room temperature it conducts poorly, and there is no way to control it. The entire semiconductor industry rests on the observation that adding a very small, very precise amount of the wrong element changes that completely. The numbers are worth pausing on: a typical doping level is one foreign atom per hundred million silicon atoms, and that is enough to change conductivity by orders of magnitude.
n-Type and p-Type
Silicon has four valence electrons, each committed to a bond. Substitute an atom with five valence electrons - phosphorus, arsenic or antimony - and four go into bonds while the fifth has nowhere to sit. It is bound only weakly and is freed by ordinary room-temperature thermal energy. The material now has mobile negative carriers and is called n-type.
Substitute an atom with three valence electrons instead - boron, aluminium or gallium - and one bond is left short. A neighbouring electron can move in to complete it, leaving its own bond incomplete, and that vacancy migrates through the crystal. The material has mobile positive carriers and is called p-type.
In both cases the crystal remains electrically neutral overall. An n-type region has extra mobile electrons but also extra fixed positive nuclei to match; nothing has been charged, only made mobile. This is a point worth being clear about, because it is easy to picture doping as adding charge, and it does not.
The dopant concentration sets the carrier concentration directly, which is why doping is such a precise engineering tool. Light doping gives high-resistivity material; heavy doping approaches metallic behaviour. Modern processes place dopants by ion implantation, firing them into the crystal at controlled energy and then annealing to repair the damage.
The Junction Is the Point
A doped region on its own is just a resistor with a chosen resistance. Everything interesting happens where n-type meets p-type.
At that boundary, electrons from the n-side diffuse across into the p-side and holes go the other way, and each leaves behind a fixed charged atom it can no longer neutralise. A region depleted of mobile carriers builds up, containing fixed positive charge on one side and fixed negative on the other. That separated charge creates an electric field, and the field opposes further diffusion until equilibrium is reached.
This built-in field is the working part of a diode, a transistor and a solar cell. When light creates an electron-hole pair inside or near that field, the pair is pulled apart before it can recombine, and the separated carriers can be collected as current. That is the photovoltaic mechanism in one sentence.
It is worth stating plainly what the built-in field does not do. It exists at equilibrium and does no net work. Connect a p-n junction to a wire in the dark and no current flows, because the same physics that creates the field also cancels it around the circuit. Energy has to come in from outside - a photon, an applied voltage - before anything useful happens.
Doping Materials Without a Band Gap
The picture above assumes a band gap with empty states above it, and graphene has none. Doping it therefore means something different: rather than filling states in an empty band, it shifts the Fermi level away from the point where the bands touch, so that the material carries more electrons or more holes.
This can be done chemically, by adsorbing molecules onto the surface that donate or withdraw electrons, or electrostatically, by applying a voltage through a nearby gate. Because graphene is a single atomic layer, everything it touches acts as a dopant to some degree - which makes it extraordinarily sensitive as a sensor and extraordinarily awkward to keep stable.
This sensitivity is a genuine engineering constraint in any layered device. A graphene layer's doping level depends on what is above and below it, on adsorbed contaminants, and on the substrate. Two nominally identical stacks can behave differently, and reproducibility is a real problem rather than a detail.
For a stack alternating doped graphene with silicon, such as the one our research concerns, the doping of each graphene layer sets its work function, which in turn sets the built-in field at the interface with the neighbouring silicon. That is the design lever. What it produces has to be established by measuring the finished device, because the interfaces, not the ingredients, determine the outcome.
Frequently asked questions
How much dopant is actually added?
Typically between one foreign atom per hundred million host atoms and one per thousand. Even at the low end that is enough to change conductivity by several orders of magnitude.
What is the difference between n-type and p-type?
n-type uses atoms with one more valence electron than the host, supplying mobile electrons. p-type uses atoms with one fewer, creating mobile holes. Both remain electrically neutral overall.
Does doping add energy to the material?
No. It changes where the Fermi level sits and how many mobile carriers exist. The material is still neutral, and no energy has been stored by doping it.
Why does the p-n junction matter so much?
Because the built-in field at the boundary is what separates an electron-hole pair before it recombines. That is the operating principle of diodes, transistors and solar cells alike.
How is graphene doped if it has no band gap?
By shifting its Fermi level away from the point where its bands touch, either chemically through adsorbed molecules or electrostatically through a gate. Because it is one atom thick, everything it touches dopes it to some degree.
