This Small Business Innovation Research Phase I project proposes to demonstrate a Composite Resonator VCSEL (CRVCL) which will achieve a 40-100Gbps data transmission rate. For more than a decade VCSELs have been the engine driving bandwidth increases at shorter distances (<300 meters) between cabinets in high performance computers, in local area networks (LANs) and storage area networks (SANs), but it appears that conventional VCSEL technology may be running out of steam somewhere between 20 and 30Gbps. The proposed approach is unique in that it avoids the relaxation oscillation limitations on modulation speed through the use of a dual cavity VCSEL operated in a push-pull configuration. The concept dramatically increases the data rate by keeping the current density constant while modulating the light to be emitted from the surface or directed to the substrate. The goals of the Phase I project are to fabricate a CRVCL, demonstrate the push-pull nature of the modulation to a minimum of 20GHz, and demonstrate the elimination or dramatic reduction in relaxation oscillations. If successful, the demonstration of the 40-100Gbps large signal modulation will be proposed for a Phase II project.
The broader impact/ commercial potential of this project is to enable the continued expansion of bandwidth within LANs, SANs and between cabinets of high performance computers. In addition, a 40Gbps+ VCSEL will be a breakthrough technology for overcoming the interconnect bottlenecks between boards within a cabinet, and within a board. Copper based interconnects are increasingly the limiting factor in system performance due to their size and power consumption, but the replacement by optical interconnects will require high speed per channel (40Gbps), very low power, high reliability, and low cost. This project concept addresses all four considerations. Higher speed devices feed the bandwidth expansion allowing the continued improvements in business productivity, management of medical information, and entertainment options we have come to expect, and helps maintain the U.S. competitiveness and employment in the networking market. The reduction in power consumption helps achieve the industry and government goals for reducing the power consumption of data centers. The successful development of this technological enhancement will help the U.S. maintain its leading position in the computing and networking industries.
The goal of this Small Business Innovation Research Phase I project was to demonstrate the feasibility of building a very high data rate (40-100 Gigabit per second) VCSEL using a Composite Resonator VCSEL (CRVCL) design. The continued rapid increase in Internet bandwidth demand requires higher speed data links at all distances, including within and between cabinets in servers and high performance computing, and even between and on boards within a cabinet. In addition to allowing continued expansion of bandwidth in Storage Area Networks or High Performance Computing, a 40Gbps+ VCSEL will be essential for the development of optical interconnects within a cabinet or circuit board. Copper based interconnects are increasingly the limiting factor in system performance due to their size and power consumption, but the replacement by optical interconnects will require high speed per channel (40Gbps), very low power, high reliability, and low cost. Our concept addresses all four considerations. Higher speed devices feed the bandwidth expansion allowing the continued improvements in business productivity, management of medical information, and entertainment options we have come to expect, and helps maintain the U.S. competitiveness and employment in the networking market. The reduction in power consumption helps achieve the industry and government goals for reducing the power consumption of data centers. VCSELs have been the engine driving bandwidth increases at shorter distances (<300 meters) for more than a decade, but it appears that conventional VCSEL technology may be running out of steam somewhere between 20 and 30Gbps. Straightforward design enhancement of conventional VCSELs will not satisfy the requirements. The novelty of the design proposed in this project has been validated by the issuance of one patent and the allowance of claims in a second patent application. The approach avoids relaxation oscillation limitations on modulation speed through the use of a dual cavity VCSEL operated in a push-pull configuration. Extensive modeling of the small signal and large signal modulation of the device was performed, which predicted the ability to achieve large signal modulation to greater than 80 Gigabits per second, achieved at a current level equivalent to today’s 10Gigabit per second devices. The modeling also allowed us to specify the parameters important to the design, such as the optimum reflectivity of the mirrors and aperture diameter. While the design approach uses the same manufacturing infrastructure currently used for conventional VCSELs, the epitaxial structure was more complex and the fabrication into a coupled cavity VCSEL structure required significant process development. For a conventional VCSEL, single crystal layers of a semiconductor material are grown on a substrate to form two mirrors with an active region sandwiched between the two mirrors. The active region is the location of the p-n junction where injected current combines and emits light. For a coupled cavity VCSEL, two active layers are grown, sandwiched between 3 mirrors. One key outcome was the successful demonstration that these layers could be grown with the very precise (<0.5%) thicknesses required to make sure the cavities are matched in thickness and hence are properly coupled to each other. A conventional VCSEL also requires etching a trench and oxidizing a layer in the structure to control the current flow through the device. The coupled cavity structure requires etching several times, and oxidizing two layers, as well as forming an electrical contact in the middle of the device. In this project we experienced, and solved, several challenges associated with multiple etches and two oxidations. We started with one process flow, but subsequently developed a new process flow that was both simpler and more successful. We were able to demonstrate the independent lasing of the two cavities in the structure, demonstrating the quality of the epitaxial structure, the ability to make electrical contact to all three terminals of the device, and the low power operation that we expect. Therefore, the following outcomes of the project helped to establish the feasibility of creating a very high speed coupled cavity VCSEL: a) the appropriate design parameters for the epitaxial growth and mask design for fabricating the VCSEL were verified by modeling, b) the ability to grow the epitaxial layers per the design was demonstrated, c) the designed fabrication process was tested and then modified for improved simplicity and yield, and d) the design of the epitaxy and process was verified by demonstrating lasing in both cavities of the structure.
