The amount of data collected, transferred and processed increased over the last decade with a substantial annual rate. With the internet of things in sight, this trend will be continued. Already today more than 10% of the electrical power produced in the western world is consumed by IT products, underlining the demand to drastically reduce the power consumption. In fact, energy consumption will be the main driver of development in information technology. Silicon photonics is considered to be an attractive pathway to drastically reduce power consumption in information technology based on classical Si microelectronics /1/. Moreover Si photonics are envisioned also as a promising pathway towards quantum computing/2/. Group IV alloys, namely Ge and SiGe on Si, are heavily investigated for the integration of modulators and detectors on Si. However, higher integration levels of optical data transfer on the chip will require the integration of low power laser. The latter is approached by a wide range of technologies to deposit III/V structures on Si as well as by bonding III/V lasers Si. Recently, a direct bandgap as well as optically pumped lasing was demonstrated for GeSn alloys deposited on Ge virtual substrates on Si (100) 8” wafers. Here, efforts are reported of this ongoing endeavor towards an electrically pumped group IV laser. The group IV alloy GeSn provide a direct band gap for Sn concentration above ~8%, which is far beyond the solid solubility limit of ~1%. Recently high quality GeSn alloys with Sn concentrations up to 14.5% could be grown at low temperatures and high growth rates by means reactive gas source epitaxy. These GeSn have a direct band gap, in the range of 0.48-0.63 eV dependening on the Sn concentration. Phosphorous and Boron doping has been realized to fabricate p-i-n junction in SiGeSn/GeSn structures. Thus these materials pave the road for the integration of optoelectronic circuitry on Si (100) substrates. Optically pumped laser in the Fabry Perot geometry as well as microdisc lasers have been fabricated from GeSn films, SiGeSn double heterostructures and SiGeSn/GeSn multiple quantum wells (MQW) on Ge virtual substrates. The threshold required to achieve lasing dropped from ~300 kW/cm2 for a thick GeSn film to about 30 kW/cm2 for a MQW structure. This can be attributed to reduced optical losses due to surface scattering as well as due to the reduced number of states in the multiple quantum wells, which allows carrier inversion at smaller pumping powers. However, due to the very small effective mass of electrons in the Г valley compared to large mass in the L valley, the quantum wells have to be designed carefully. For thin quantum wells the confinement shift of the subbands of the Г valley is larger than that for the L valley reducing the directness and may even turn it into an indirect material. Moreover, a (Si)GeSn buffer layer above critical thickness is required to relax the strain before the growth of the active region of the laser. The strain relaxation occurs predominantly via 90° step edge dislocations at the Ge/GeSn interface, within the hetero- and MQW structures no dislocations were detected by cross sectional TEM. However, besides defect engineering, care has to be taken in the design of the structures also to avoid losses due to absorption in GeSn and SiGeSn cladding layers, since the limited band offsets give rather stringent design rules. To achieve electrically pumped lasing double hetero- and MQW-structures have been grown using SiGeSn cladding and barrier layers. Parts of the SiGeSn cladding layers have been doped by P and B to achieve p-i-n junctions. First devices have been fabricated showing a superior electroluminescence efficiency of MQW structures compared to hetero- of homojunction devices. /1/ R. Soref, IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, 12, 1678 (2006) /2/ T. Rudolph APL PHOTONICS 2, 030901 (2017)
Read full abstract