Optical technologies have advanced dramatically in recent years. In just two decades the transparency of optical fibers has improved by four orders of magnitude. Semiconductor lasers have evolved from a new invention to highly reliable, high performance commercial devices for wide bandwidth optical communications. New approaches to higher frequency modulation, wider bandwidth transmission, more sensitive detection and optical amplification are constantly being developed. Fundamental limitations are sufficiently far removed from current capabilities that considerable further progress can be anticipated. These advances have provided the stimulus for a much broader investigation of the potential of optics in future information technologies in which optics and electronics play complementary roles. This rapidly developing field is referred to as “photonics.” Increasing attention is now being paid to applying optics to wide bandwidth switching systems and to exploring the potential of optics for image processing and computation.

Past progress in optical communication can be traced largely to the dramatic progress in optical fiber and compound semiconductor materials technologies. Likewise, future opportunities in photonic switching and information processing will depend critically on the development of improved photonic materials. The future role of optics in these conventionally electronic technologies, and the extent of that role, depends on whether materials can be designed and fabricated with the required characteristics.

In electronic processors, heat dissipation and interconnection delay are serious design and performance limiting factors. Why then consider photonic components where both the energy and size of the photon are large (˜1 eV and ˜1 μm, respectively) and the required nonlinear interactions between electric or magnetic fields and photons for switching or modulation are small? There are several answers to this question. First, the wide bandwidth of optical communications systems is taxing the current capabilities of electronic switching technologies. Even a slow optical switch can switch a very wide bandwidth optical signal from one fiber to another. External optical modulators will likely be required in ultrawide bandwidth communications because of basic limitations on direct modulation of lasers. Because of the weak electromagnetic interaction and low dispersion, optical interconnection of electronic circuits offers considerable advantages in high speed computer architectures. Some of these applications would appear to be relatively near term since they build on current capabilities of optical communication.

Longer term and more speculative are applications of photonics to computation and image processing — areas where electronics technology is already mature. Current research can be divided into two groups — ultrafast processing and parallel processing. The first group concentrates on processing with ultra-fast optical pulses. Optical pulses as short as 6 fs — orders of magnitude shorter than any electronic pulses — have been generated in the research laboratory. High processing rates are achievable by serial processing of high repetition rate ultrashort pulses. This approach requires ultrafast switches, which in turn requires materials with ultrafast nonlinear optical response time. Indeed, the shortest electrical signals are now measured by optical sampling techniques.

Planar optical-guided-wave devices have been in existence for over 20 years. Two interesting, informative articles can be found in References 1 and 2. Much of the early work in guided-wave optics was on passive devices, but this was also when much of the theoretical understanding of optical-guided-wave (OGW) devices was developed. This theoretical understanding applies to active devices as well. It's interesting to note that the commercialization of passive, glass-based guided-wave devices has just occurred with product introductions by Corning & Nippon Sheet Glass. Active OGW devices (i.e., ones where the light properties can be altered with an applied voltage) have been reported since about 1975. In 1985, Crystal Technology, Inc. announced the first commercially available product—a high-speed, efficient, intensity modulator. Throughout the ten years in between, a huge amount of technical literature has been generated. Most of the work has centered on the Ti:LiNbO3 waveguide technology although several other material systems have been demonstrated as well.

Materials upon which optical-guided-wave modulators have been fabricated include: dielectrics such as lithium niobate (LiNbO3), lithium tantalate (LiTaO3), and potassium titanyl phosphate (KTP); the III-V semiconductor compounds, gallium arsenide (GaAs) and indium phosphide (InP); and a variety of organic polymers. Of these materials, waveguides on LiNbO3 are clearly the most developed and are offered for sale commercially. For this reason I will concentrate on this material system while making comparisons to the other material systems when appropriate.

were very exciting but speculative, being technologically feasible only if new classes of materials could be developed The subject of materials in nonlinear optics (NLO) encompasses a wide range of important topics. Today the line between materials and NLO processes has become fuzzy, particularly for newer NLO processes (e.g. photorefrac-tion, and optical bistability, logic and computing). For more established NLO processes (e.g., harmonic generation, parametric processes, linear electro-optic effect, etc.) the subjects are well studied and the importance of various materials properties on the NLO process are known, though these properties are not necessarily predictable, controllable, or optimized in current materials.

A decade ago, having been introduced to NLO phenomena through postdoctoral research, I had an opportunity to define and pursue an NLO research program at Xerox's Webster Research Center. The question was posed: “Are new materials needed for NLO applications?” The answer must start with another question: “Which NLO process … with light of what wavelength, pulse duration, and power… and for what purpose?”

It was clear that important limitations to many of the novel things one might do with optics were: insufficient nonlin-earity magnitude, inability to fabricate reliable device structures, occurrence of deleterious optical properties, and restrictions due to other material properties. The newer NLO phenomena. Use of older NLO processes in new technological applications seemed a more down-to-earth quest.

From the time of their conception, holography and holograms have evolved as a metaphor for human memory. Holograms can be made so that the information they contain is distributed throughout the holographic medium—destroy part of the hologram and the stored information remains wholly intact, except for a loss of detail. In this property holograms evidently have something in common with human memory, which is to some extent resilient against physical damage to the brain. There is much more to the metaphor than simply that information is stored in a distributed manner.

Research in the optics community is now looking to holography, in particular dynamic holography, not only for information storage, but for information processing as well. The ideas are based upon neural network models. Neural networks are models for processing that are inspired by the apparent architecture of the brain. This is a processing paradigm that is new to optics. From within this network paradigm we look to build machines that can store and recall information associatively, play back a chain of recorded events, undergo learning and possibly forgetting, make decisions, adapt to a particular environment, and self-organize to evolve some desirable behavior. We hope that neural network models will give rise to optical machines for memory, speech processing, visual processing, language acquisition, motor control, and so on.

Over the past four decades, the growth of information processing and computational capacity has been truly remarkable, paced to a large extent by equally remarkable progress in the integration and ultra-miniaturization of semiconductor devices. And yet it is becoming increasingly apparent that currently envisioned electronic processors and computers are rapidly approaching technological barriers that delimit processing speed, computational sophistication, and throughput per unit dissipated power. This realization has in turn led to intensive efforts to circumvent such bottlenecks through appropriate advances in processor architecture, multiprocessor distributed tasking, and software-defined algorithms.

An alternative strategy that may yield significant computational enhancements for certain broad classes of problems involves the utilization of multidimensional optical components capable of modulating and/or redirecting information-carrying light wave-fronts. Such an optical processing or computing approach relies for its competitive advantage principally on massive parallelism in conjunction with relative ease of implementation of complex (weighted) interconnections among many (perhaps simple) processing elements. A wide range of computational problems exist that lend themselves quite naturally to optical processing architectures, including pattern recognition, earth resources data acquisition and analysis, texture discrimination, synthetic aperture radar (SAR) image formation, radar ambiguity function generation, spread spectrum identification and analysis, systolic array processing, phased array beam steering, and artificial (robotic) vision.