Getting photonic crystal nano-lasers on silicon

Recent demonstrations of photonic crystal lasers on a silicon platform highlight the tremendous potential of these devices for providing efficient light sources for silicon nanophotonic integrated circuits.

BY MINGCHU TANG FROM UNIVERSITY COLLEGE LONDON

Semiconductor lasers have come an awfully long way since they emerged from a number of industrial labs in the US in 1962. To realise lasing in those first homostructure devices, the chips would be cooled by liquid nitrogen and electrically pumped with incredibly short, high-current pulses. Fortunately, rapid progress followed the invention of this device, with the introduction of double heterostructures and refinements to the active region improving key characteristics and enabling the production of a practical device.

These advances have spurred the commercialisation of the semiconductor laser and its widespread deployment. Significant successes include miniature sources for reading optical discs, the key technology for CD and DVD players, and the manufacture of countless light engines for various optical networks that underpin communication and the internet.

Now there is much interest in using III-V lasers for silicon photonics. Producing integrated circuits that route light through waveguides and a number of on-chip devices enables energy savings in many different applications requiring data transfer, such as high-performance computing and data centres, as well as opening up many new markets, including those in healthcare, where miniaturisation is highly valued.

A significant challenge with any silicon photonic integrated circuit is how to incorporate the III-V laser onto the chip. One option is bonding, using the likes of benzocyclobutene (BCB) or oxygen atoms to form a bonding interface. Intel has successfully commercialised this integration method, employing it for the production of silicon optical transceivers operating at 100 Gbit/s. However, the price of these components is at odds with the needs of silicon photonics, which requires a low-cost, high-yield, CMOS-compatible optical communication platform.

A more attractive alternative for bringing laser light to the silicon photonic chip is direct epitaxy. MBE and MOCVD have been used to grow III-Vs on silicon platforms. Recently, this has been shown to be an efficient way to fabricate silicon-based III-V lasers, due to the merits of low cost and large scale. However, despite significant demonstration of conventional Fabry-Pérot and distributed feedback lasers on III-V and silicon platforms, there is still the need for microscale and nanoscale laser devices with far lower energy consumption and optical mode control for silicon-based nanophotonic integrated circuits. Such circuits are promising candidates for next-generation quantum computing and optical microprocessors, and could be used to make microelectronic components for optical/photonic microprocessors.

Helping to lay the foundations for the development of miniature laser sources are the microdisk lasers, a triumph of the 1990s. At the heart of these devices are micro-resonators, a few micrometres in diameter. These structures support whispering-gallery modes, instrumental to single-mode lasing and low operating powers.

Photonic crystal cavities
Building on this concept – and shrinking the footprint of the laser while maintaining its excellent performance – are designs employing a photonic crystal cavity. Thanks to an enhanced light-matter interaction within the photonic cavity, the photonic crystal laser not only benefits from a smaller footprint that is nanoscale in size, but strong optical confinement that comes from the designed photonic bandgap (the primary strengths of the photonic crystal laser, compared with its edge-emitting variant, are listed in Table 1).