YIP: Optoelectronics: Ultrafast Directly Modulated NanoLED for On-chip Optical Interconnect
Abstract
The microelectronics industry has embraced parallelism to sustain performance growth in the last decade. However, the benefits of parallelism are constrained not by computation capability at individual nodes, but by data movement speed between nodes. In order to increase data transfer rate by implementing optical interconnects, there is an urgent need to drastically improve the interconnectivity between the photonic ÒplaneÓ and electronic ÒplaneÓ for multi-functional adaptive photonic/electronic integration. To achieve this, three grand challenges Ð integration density, speed, and energy efficiency Ð have to be addressed. In the search for the next generation on-chip light source that can address all of the above challenges, nanoscale LEDs, rather than the more conventional choice of (nano)lasers, were proposed as the ideal candidate in several recent studies. We share this vision and propose a novel and comprehensive nanoLED design and analysis approach to overcome the currently low efficiency in nanoscale light sources. Complemented with experimental validation, this work will enable the continued scaling of information processing and communication systems. The objective of this project is to investigate nanoLED performance at both the device and system level by coupling nanoLED cavity and driver circuitry designs for the first time, in order to comprehensively analyze the system efficiency and to show the viability of nanoLED-enabled optical interconnects. Through the multi-physics consideration of optical, electrical and thermal performance, combined with wall-plug efficiency evaluation, we optimize the nanoLED design not only for superior emission characteristics through novel cavity design, but also for optimal overall system-level performance by including driver circuitry on-chip. Experimentally, we measure the device speed under both small- and large-signal modulations. The proposed work can enable a host of applications that is of interest to the military in the realm of big data, including machine learning, large scale emulation, and high-performance computing. Additional DoD applications that will benefit from this research include advanced sensors, wireless interfaces, LIDAR image analysis, target identification, and RF photonics.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 10, 2019
- Source ID
- W911NF1910303
Entities
People
- Qing Gu
Organizations
- Army Contracting Command
- United States Army
- University of Texas at Dallas