Silicon has many advantages that have made it the dominant semiconductor material:
- A native oxide. This is key to the development of the MOSFET.
- Relatively good physical robustness. Some other competing materials are more fragile, leading to losses in production simply due to mechanical breakage of wafers.
- Abundance in nature. Silicon is the 2nd most abundant element in the Earth's crust, making it easy to mine, though refining it to the purity needed for electronics is still a significant effort.
Further, since silicon is so widely used, economies of scale make it much cheaper to produce chips or devices in silicon than in other semiconductors.
So, if silicon will do the job, we will almost always choose silicon to achieve low cost.
We might choose other materials if we need
- A direct-gap material, typically for optical sources like LEDs.
- A specific bandgap. For example for photodiodes for detecting 1550 nm wavelength, a bandgap less than about 0.8 eV is needed.
- High carrier mobility, which allows higher frequency devices. For this you'll see materials like SiGe, GaAs, GaN, or InP used.
- A specific lattice constant, for growing one material epitaxially on a substrate of another material. The ability to engineer both lattice constant and bandgap is why you see ternary and quaternary compounds like GaAlAsP used.
I'll leave aside the question of how dopants are chosen because 1) I know almost nothing about it, and 2) the choice of dopants is likely different for each semiconductor material.
We are not close to making pure Ge ICs, and likely never will be. The odd diode or power device where the low voltage wins, yes, but the process technology for significant integration, no.
The technology to be able to process silicon into ICs as we do today has been won over 40 to 60 years (depending on your start date) of painstaking incremental investment. Given a choice between a tweak that uses the existing silicon processes (continue to reduce dimensions, alloy Ge with it, put it on an insulating sapphire substrate) and a totally new material to learn how to process and handle the chemistry, the choice will be for silicon, unless there is some huge advantage for the new material.
A big, big drawback for straight Ge is the low temperature performance of the material. You can't solder ICs by bringing the whole board up to temperature as you can with silicon, it's hand solder leads or clamp only, and then use excellent heatsinking. What the market wants in higher temperature performance, that's one of the things driving SiC for instance.
IC require much more than discretes, every aspect of the physics and chemistry, the resistors, the well capacitance, the oxide stability, the interconnect metals, the lattice sizes to mention a few from a long list, all form part of a system within which the IC designers work.
Discretes are a little more forgiving, and here new materials do get used at a low level of integration. GaN is making higher speed FETs, SiC is making higher voltage higher temperature transistors, SiGe first appeared in discretes for a decade or two where the teething troubles were sorted out, then there's InGaP, and AlGaAs, and a host of others, and Ge too for some specialist diodes and power transistors.
Given how long Ge has been around, it's obvious that there is not a sufficient performance advantage to make it worth trying to work with the material's problems on any significant scale.
Best Answer
SiGe is a semiconductor alloy, meaning a mixture of two elements, silicon and germanium. Since 2000 or so, SiGe has become widely used to enhance the performance of ICs of various types. SiGe can be processed on equipment nearly the same as used for ordinary silicon. SiGe doesn't have some of the drawbacks of III-V compound semiconductors like gallium arsenide (GaAs), for example it doesn't lack a native oxide (important for forming MOS structures) and doesn't suffer from mechanical fragility that limits the wafer size of GaAs. This results in costs that are only a small multiple of ordinary silicon, and so much lower than competing technologies like GaAs.
SiGe allows two main improvements compared to ordinary silicon:
First, adding germanium increases the lattice constant of the alloy. If a layer of Si is grown on top of SiGe, there will be mechanical strain induced by the lattice constant mismatch. The strained layer will have higher carrier mobility than unstrained Si. This can be used, for example, to balance the performance of PMOS and NMOS transistors, reducing the area needed for a given CMOS circuit.
Second, the SiGe alloy can be used selectively in the base region of a BJT to form a heterojunction bipolar transistor (HBT). SiGe HBT's have been demonstrated with speeds (fT) to 500 GHz, and are commercially available with fT up to 240 GHz. The SiGe HBT also has lower noise than a standard silicon BJT.