Figure 1 | Strained germanium nanowire lasers. a, Schematic illustration of a typical germanium nanowire laser consisting of a strained nanowire surrounded by a pair of DBRs on the stressing pads. Top inset: SEM image. Scale bar, 10 µm. Bottom inset: TEM image of the GOI structure. Scale bar, 0.5 µm. b, Top: top-view SEM image. Scale bar, 5 µm. Middle: 2D strain map. Bottom: simulated optical field distribution. c, Calculated bandstructure of 1.6% uniaxial strained Ge.
Photonic-integrated circuits (PICs) are a key enabler for the ultimate miniaturization of various disruptive technologies such as LiDAR for autonomous vehicles, bio-chemical sensors and chip-level optical communications. However, the realization of fully functional PICs is currently limited by the absence of an efficient laser source on silicon (Si), thus making a practical Si-compatible laser the Holy Grail. It has been thus far considered close to impossible to create a practical on-chip laser using Si-compatible group IV semiconductor materials (e.g. germanium (Ge)) owing to their indirect bandgap nature.
In the quest for practical group IV lasers, researchers have proposed a few innovative ideas such as strain engineering of Ge and alloying of tin (Sn) into Ge. The strain engineering of Ge holds a clear advantage over the Sn alloying since it bypasses the formidable task of the material growth. Over the past few years, various creative platforms for strain engineering have been experimentally demonstrated (Nature Photon. 6, 398–405 (2012) / Nature Photon. 7, 466–472 (2013)). However, despite several years of attempts, there has been no successful demonstration of lasing in highly strained Ge. Our group recently reported the first experimental observation of low-threshold lasing in highly strained Ge nanowire at 83 K. The lasing threshold is nearly two orders of magnitude lower than the state-of-the-art GeSn laser. This striking achievement is strongly supported by a rigorous theoretical simulation, which predicts a large optical net gain in strained Ge. By presenting unambiguous quantitative evidence of low-threshold, multimode lasing action in a fully CMOS-compatible material system, our results pave the way towards a truly monolithic realization of PICs.
Figure 2 | Strained-induced pseudoheterostructure. a, SEM image of a fabricated and fully suspended structure. Scale bar, 5 μm. b, Schematic of the energy band diagram along the center of the structure showing a strain-induced potential well and captured carriers. c, 2D PL map of a strain-induced double heterostructure, showing bright emission along the active region. d, 2D PL map of a strain-induced graded double heterostructure, showing concentrated, bright emission only at the center of the active region.
The advent of “quantum heterostructures”, which led to the 2000 Nobel Prize in Physics, has enabled a host of essential devices from modern lasers to tandem solar cells and beyond. Research in the field of quantum heterostructures still remains very active: quantum wells and quantum dots, for instances, are all specific examples of quantum heterostructures. Using a so-called “quantum heterostructure” wherein a narrow band gap material with thicknesses in quantum regime is sandwiched between two wide band gap materials, charge carriers can readily be confined in the narrow band gap material, thereby generating various exciting quantum phenomenon.
Despite their critical role in modern device technology, integrating different semiconductor materials for quantum heterostructures still presents significant barriers; heteroepitaxy is costly due to the associated chemical vapor deposition (CVD) or molecular beam epitaxy (MBE) processes and often further complicated by large lattice mismatches and/or incompatible thermal budgets for different semiconductors. Moreover, since a different process step is required for each new material, only a small number of unique quantum heterojunctions can be fabricated on a single wafer. All of these problems can be avoided if one can create quantum heterostructures within a single material by locally modifying the material’s band gap. Strain engineering is a well-established technique to modify the electronic band gap of virtually all semiconductors. The lattice constant of a crystalline semiconductor increases with tensile strain, thereby reducing its electronic band gap.
Our research team has invented a fundamentally new route towards creating quantum heterostructure in group-IV materials (e.g. germanium) and 2D materials (e.g. MoS2) only by leveraging spatially varying strain enabled by a conventional, simple lithography. We coin the term Q-SET, Quantum Strain Engineering Technology, for this strikingly new technology. A major advantage of Q-SET is that the properties of the quantum heterostructures (e.g. their sizes and band offsets) depend only on the device geometry defined by lithography, rather than variations in the material composition. The inherent flexibility of our technique also allows for the generation of an unprecedented number of unique quantum heterostructures on a single chip.
Figure 3 | Band structure engineered germanium-tin lasers. a, Color mapping of the bandgap as a function of biaxial tensile strain and tin concentration. b, Top: SEM image of a fully released germanium-tin photonic crystal laser structure with >8% tin content. Left bottom: simulated optical field distribution. Right bottom: 2D strain map.
Alloying germanium with tin has recently attracted much attention for photonic-integrated circuits because germanium-tin with a reasonably high tin concentration ( >6.5%) can achieve the direct bandgap, thereby enabling an efficient silicon-compatible light source. Also, its bandgap energy can be reduced sufficiently to detect the mid-infrared ( >2000 nm) light. According to recent theoretical calculations, germanium-tin with ~10% tin concentration demonstrates a direct band-gap nature, thus showing great promise as a potential lasing medium.
In fact, Wirths et al. recently demonstrated optically pumped lasing in germanium-tin alloy at cryogenic temperature (Nature Photon. 9, 88–92 (2015)). However, due to the high threshold ( >300 kW/cm²) and a low-temperature operation (< 100 K) for reported germanium-tin lasers, the feasibility of practical group-IV lasers still remains elusive.
Via rigorous theoretical modelling and high-quality nanofabrication, our team is currently developing truly practical germanium-tin lasers with a lower threshold level (1-10 kW/cm²) and an increased operating temperature to be used with a convenient thermo-electric cooler (200 K and above) while the ultimate goal is to achieve room temperature operation.