1.3 µm Quantum Dot‐Distributed Feedback Lasers Directly Grown on (001) Si

Abstract

Distributed feedback (DFB) lasers represent a central focus for wavelength‐division‐multiplexing‐based transceivers in metropolitan networks. Here, the first 1.3 µm quantum dot (QD) DFB lasers grown on a complementary metal‐oxide‐semiconductor (CMOS)‐compatible (001) Si substrate are reported. Temperature‐stable, single‐longitudinal‐mode operation is achieved with a side‐mode suppression ratio of more than 50 dB and a threshold current density of 440 A cm−2. A single‐lane rate of 128 Gbit s−1 with a net spectral efficiency of 1.67 bits−1 Hz−1 is demonstrated, with an aggregate total transmission capacity of 640 Gbit s−1 using five channels in the O‐band. Apart from the QD active region growth, the overall fabrication is essentially identical to the commercial process for quantum well (QW) DFB lasers. This provides a process‐compatible path for QD technology into commercial applications previously filled by QW devices. In addition, the capability to grow laser epi across entire CMOS‐compatible (001) Si wafers adds extra benefits of reduced cost, improved heat dissipation, and manufacturing scalability. Through direct epitaxial integration of III–Vs and Si, one can envision a revolution of the photonics industry in the same way that CMOS design and processing revolutionize the microelectronics industry. This is discussed from a system perspective for on‐chip optical interconnects.

High Speed Evanescent Quantum‐Dot Lasers on Si

“…A maximum SMSR of more than 60 dB is achieved at an injection current of 70 mA, much higher than that of the best reported QD‐based DFB lasers, which fall into the range of 27–47 dB. [ 12,22,31 ]…”

“…Externally modulated QD DFB lasers as optical sources are attractive and several integration schemes to integrate QD DFBs and electroabsorption modulators (EAMs) have been explored. [ 12 ] While terabit Si transmitters utilizing heteroepitaxial integration is possible, modulators inherently waste at least 3 dB (6 dB common) in power consumption and the on‐chip coupling costs an additional 1–2 dB of loss. Therefore, higher power efficiency is an advantage for directly modulated lasers, especially for the 10 Gb s −1 × 10 and 25 Gb s −1 × 4 lane architectures that demand inexpensive low‐bandwidth lasers.…”

“…Regrown QD distributed feedback (DFB) lasers have demonstrated a process‐compatible path for QD technology into commercial applications previously filled by QW devices. [ 12 ] A monolithic offset QD integration platform has been established enabling the formation of a laser cavity utilizing both a robust QD active region and versatile passive GaAs waveguide structures. [ 13 ] The aforementioned two techniques are similar to the case of photonic integration on InP, except that the III–V layers are grown on a cheaper and larger‐diameter Si substrate.…”

An advanced III-V-on-silicon photonic integration platform

“…As it's potentially to be the ultimate solution for very high density integration, it has regained huge interest 12−17 . As a result of newly developed novel substrate patterns, intermediate buffer layers, and the use of defect-tolerant active regions, e.g., quantum dot (QD), significant progress has been achieved with the demonstrations of lasers with high efficiency 13 , low threshold 14 , direct modulation 16 and good lifetime 18 . However, several other critical issues associated with this approach currently include potentially more challenges in heteroepitaxy on a SOI substrate over bulk Si ones; difficulty in achieving efficient light coupling from the III/V active region to the Si waveguide due to inevitable several μm-thick buffer layers; and extra optical loss when light propagates in the dense dislocation zones.…”

Monolithic integration of InP on Si by molten alloy driven selective area epitaxial growth

“…In recent years tremendous effort has been devoted to implementing techniques that reduce the number of defects in the III-V nanostructures monolithically integrated into Si. They include wafer bonding [14][15][16][17] , transfer printing 18 and direct epitaxial methods of integration 13,19 , such as growth on thick buffer layers [20][21][22] , various selective-area growth (SAG) methods 12,[23][24][25][26][27] and different approaches utilizing liquid group III metallic droplet as a growth catalyst [28][29][30] . The latter two families of techniques are more promising, due to their flexible integration into Si-based circuits and avoiding usage of III-V substrates, complicated wafer bonding techniques and difficulties in alignment of III-V elements to the Si-circuit.…”