Silicon photonics technology is becoming more and more popular: experience the progress of silicon light emitting technology (Part 3)

Reducing modulator size through “slow light”

　　To further improve the results of PECST, it is crucial to further reduce the size of the optical modulator and achieve high-speed operation. Research in this area has also made progress (Figure 7). For example, the research and development team of Professor Toshihiko Baba of the Graduate School of Engineering of Yokohama National University, one of the researchers of PECST, has developed a Mach-Zehnder type optical modulator that can achieve 10Gbit/second operation using photonic crystal (PhC)* technology through CMOS compatible technology. As a result, the length of the optical modulator was greatly shortened to 90μm.

　　Figure 7: Further advances in optical modulators

　　This figure shows an overview of the next-generation optical modulator developed by a Japanese research institute. The Baba Laboratory at Yokohama National University used photonic crystals (PhC) to reduce the speed of light to about 1/10, thereby ensuring a longer effective path length of light with a shorter component length (a). The Wada Laboratory at the University of Tokyo used a combination of a germanium modulator and MEMS to successfully control the modulatable wavelength of germanium using the stress of a leaf spring (b). (Figure (a) was produced by PECST, (b) was taken by the Wada Laboratory at the University of Tokyo)

　　*Photonic Crystal (PhC) = A material in which a large number of holes with a size roughly equal to the wavelength of the electromagnetic waves passing through are artificially made in a material through which electromagnetic waves pass. It is used for light sealing, path control, group velocity control, etc. The atoms of semiconductors are arranged regularly, so carriers such as free electrons produce valence bands, forbidden bands (band gaps), and conduction bands. PhC achieves the same effect as semiconductors by replacing atoms with artificial holes. Recently, "phononic crystals" that can achieve the effect of semiconductor lattice vibrations (phonons) have also been introduced.

　　PhC is characterized by its high light confinement effect and its ability to significantly slow down the speed of light (group velocity). Slow light means that the effective refractive index of the PhC waveguide is large, and a long effective path length can be ensured even with a short waveguide, thus enabling the miniaturization of the modulator.

　　In the development of PhC, there is an example of slowing down the speed of light to about 1/10 million. However, if the speed of light is too slow, the bandwidth will be very narrow. In the development of Professor Baba, by slowing down the speed of light to about 1/10, it can be used in a 17nm bandwidth near the wavelength of 1550nm, and "the temperature dependence is relatively small, and it can operate even at temperature changes of more than 100℃."

　　According to Professor Baba, this complex structure of the component may seem difficult to manufacture at first glance, but "it can be manufactured using the 248nm KrF stepper used in 180nm process CMOS technology."

　　Importing MEMS technology

　　Another technology that is expected to reduce the size of the modulator is MEMS technology. The research and development team of Kazumi Wada, a professor at the Graduate School of Engineering at the University of Tokyo, adopted MEMS technology in the electric field absorption (EA) type modulator using germanium (Ge). As a result, the length of the modulator was reduced to about 30μm. Its characteristics are that undoped germanium can be used, and MEMS technology can also make the wavelength range used for modulation variable.

　　EA-type modulators and photodetectors using germanium generally change the modulation and light receiving wavelengths by doping or applying strain to germanium, but variable control of the wavelength cannot be achieved, and after doping, there is a problem of reduced compatibility with other components in the manufacturing process.

　　The material that was not luminous before now glows

　　The biggest remaining challenge for silicon photonics is the light-emitting element. The light-emitting elements of optical transceivers developed so far are not compatible with silicon and CMOS, so light-emitting elements using compound semiconductors must be attached. Realizing a light-emitting element compatible with CMOS can be said to be the "long-cherished wish" of silicon photonics technology.

　　Now, breakthroughs are being made in this area. Previously, silicon and germanium were considered to be semiconductors with indirect transfer band structures* and therefore basically did not emit light. However, in the past one or two years, this "common sense" has been broken, and there is hope for the realization of light-emitting elements using germanium and silicon (Figure 8).

　　Figure 8: CMOS-compatible light sources are finally becoming a reality

　　This picture shows a light-emitting element that can be made using the recently developed CMOS-compatible technology. MIT succeeded in making the Ge-on-Si element achieve laser oscillation by injecting current (a). Hitachi, Ltd. and the Arakawa Laboratory of the University of Tokyo also succeeded in making the Ge-on-Si element achieve light emission by current injection technology (b). In addition, the Otsu Laboratory of the University of Tokyo succeeded in making the pin-type silicon element achieve high-efficiency light emission (c). Light emission of multiple wavelengths has been achieved. (Figure (b) was made by PECST, (c) was taken by the Otsu Laboratory of the University of Tokyo)

　　*Indirect transfer type = When analyzing the energy band structure of a semiconductor based on wave number and electron energy, the wave number with the highest energy in the valence band is different from the wave number with the lowest energy in the conduction band. Wave number is a physical quantity related to momentum, so even if you want to transfer electrons from the conduction band to the valence band, generally speaking, it cannot transfer without complying with the law of conservation of momentum, which means that light cannot be emitted. The energy band structure that can emit light is called direct transfer type.

　　One of the research institutions that broke this common sense is the Massachusetts Institute of Technology (MIT). In 2010, MIT made germanium emit light through light excitation, and in 2012, it succeeded in making germanium realize laser oscillation by injecting electric current.

　　The secret to success is to dope germanium with high concentration n-type to change its band structure to direct migration type. The current doping concentration is 4×1019/cm3, which is very high for semiconductors. In the research on germanium, Wada of the University of Tokyo, who has communicated with MIT, confidently stated, "There is only one step left. If we can achieve doping of more than 1020/cm3, we can achieve luminescence gain equivalent to compound semiconductors. All silicon photons can be realized using Group IV materials (such as silicon and germanium)."

　　Hitachi, Ltd. and Arakawa Laboratory of the University of Tokyo have also achieved germanium luminescence. Hitachi, Ltd. has been conducting research on silicon luminescence through quantum effects until two years ago, and then began research on germanium. It also uses high-concentration n-type doping of germanium, and on this basis, applies strain to germanium through SiN, and has confirmed that this method can increase the luminescence intensity.