|
|
|
“High Throughput Optoelectronic Smart Pixel Systems Using Diffractive Optics”
by Chih-Hao Chen
May 1999
Recent developments in digital video, multimedia technology and data networks have greatly increased the demand for high bandwidth communication channels and high throughput data processing. Electronics is particularly suited for switching, amplification and logic functions, while optics is more suitable for interconnections and communications with lower energy and crosstalk. In this research, we present the design, testing, integration and demonstration of several optoelectronic smart pixel devices and system architectures. These systems integrate electronic switching/processing capability with parallel optical interconnections to provide high throughput network communication and pipeline data processing. The Smart Pixel Array Cellular Logic processor (SPARCL) is designed in 0.8_m CMOS and hybrid integrated with Multiple-Quantum-Well (MQW) devices for pipeline image processing. The Smart Pixel Network Interface (SAPIENT) is designed in 0.6_m GaAs and monolithically integrated with LEDs to implement a highly parallel optical interconnection network. The Translucent Smart Pixel Array (TRANSPAR) design is implemented in two different versions. The first version, TRANSPAR-MQW, is designed in 0.5_m CMOS and flip-chip integrated with MQW devices to provide 2-D pipeline processing and translucent networking using the Carrier-Sense-Multiple-Access/Collision-Detection (CSMA/CD) protocol. The other version, TRANSPAR-VM, is designed in 1.2_m CMOS and discretely integrated with VCSEL-MSM (Vertical-Cavity-Surface-Emitting-Laser and Metal-Semiconductor-Metal detectors) chips and driver/receiver chips on a printed circuit board. The TRANSPAR-VM provides an option of using the token ring network protocol in addition to the embedded functions of TRANSPAR-MQW. These optoelectronic smart pixel systems also require micro-optics devices to provide high resolution, high quality optical interconnections and external source arrays. In this research, we describe an innovative algorithm to design Diffractive Optical Elements (DOEs) having higher uniformity and better signal-to-noise ratio. The algorithm is based on nonlinear least-square optimization procedures and phase-shifting quantization scheme to minimize the reconstruction error of DOEs. We also describe a modified diffractive microlens design algorithm to overcome linewidth limitations in fabrication while achieving higher numerical aperture and better power efficiency. Several diffractive optical devices used in our smart pixel systems, including microlens arrays and spot array generators, are designed by these algorithms, and have been fabricated and characterized for system integration.