Why do we have to use Ge as a photodetector

Why do we have to use Ge as a photodetector
1、 Basic positioning: Why is it necessary to use Ge as a photodetector
In silicon optical links, photodetector are the “translators” who convert optical signals back into electrical signals. However, silicon itself has a bandgap of 1.12 eV and is almost transparent to 1310/1550 nm communication bands, so only germanium (Ge) can be introduced
Ge has a direct bandgap of 0.8 eV, which covers the communication O/C band, but has a 4.2% lattice mismatch with silicon. The dislocation density for direct growth is as high as 4 × 10 ⁸ cm ⁻ ², and dark current is completely unavailable; At the same time, Ge has an indirect bandgap, and its absorption coefficient is naturally one order of magnitude lower than InGaAs, which is a natural weakness.
2、 Core breakthrough: waveguide integration breaks the performance bottleneck
The “absorption length=carrier collection path” of traditional vertical incidence photodetectors has a “responsivity bandwidth” seesaw, with an upper limit of only 7GHz;
At present, the mainstream device routes are divided into three categories:
Vertical p-i-n: The process is the simplest and mainstream in the industry, achieving 40Gb/s @ zero bias and>60GHz bandwidth;
MSM Metal Semiconductor Metal: No need for high-temperature doping, can be integrated in the backend, has high dark current, and a bandwidth of over 40GHz;
High end variants: Traveling wave photodetectors (TWPD) and single line carrier photodetectors (UTC) are used for microwave photon links, balancing high bandwidth and high saturation photocurrent.
3、 Materials and Craftsmanship: Turning ‘Defects’ into Advantages
In response to lattice mismatch and performance shortcomings, the industry has developed mature solutions:
Two step epitaxy method: first, a low-temperature buffer layer of 30-50nm is grown, and then the temperature is increased to reach the target thickness, reducing the dislocation density to~10 ⁷ cm ⁻ ²;
Strain engineering: The difference in thermal expansion coefficients between Ge and Si will cause a 0.2% biaxial tensile strain in the Ge film, resulting in a direct band gap reduction from 0.8 eV to 0.77 eV and an absorption edge extension from 1.55 μ m to 1.61 μ m, covering the entire C+L band, and even the absorption coefficient in the L band can match that of InGaAs;
CMOS integration: It is still in the exploratory stage. Front end integration (FEOL) needs to withstand high temperatures above 750 ℃, while back-end integration (BEOL) is temperature friendly but without crystal substrates, and has not yet formed a unified mature solution. Currently, the industry generally adopts a mixed route of “90% single-chip+external laser“.