Frederick J. Leonberger
This chapter reviews the current status of guided wave (otherwise known as integrated optical) devices and advanced photonic packaging capability in the United States and Japan. Integrated optical devices include modulators and passive circuits in LiNbO 3, glass, and semiconductors; these devices are applied to telecommunications, cable television (CATV), and instrumentation. The photonic packaging activity discussed in this chapter is primarily related to diode lasers, but it also includes automation activities and hybrid photonic packaging.
Within the past decade, guided wave, or integrated optical (IO), components in various materials have become available from a variety of vendors worldwide and are now being deployed in commercial systems. These components are key to advanced transmitters in many fiber-optic-based CATV and long-haul telecommunications systems. The basic devices are based on planar optical waveguides, in which light is confined to channels at the substrate surface and routed on the chip. These channels are typically less than 10 microns across and are patterned using microlithography techniques. Using appropriate optical circuits based on these channel guides, both passive functions (i.e., power splitting from one to several channels) and active functions (i.e., electrical-to-optical signal conversion, known as modulation) can be performed on the light. Figure 4.1 shows a representative modulator circuit, often referred to generically as an integrated optical circuit (IOC) (Tamir 1987). The primary materials used in the commercial market are glass (bulk or SiO 2/Si) for passive devices and LiNbO 3 for active devices. A closely related area that is in the research stage is photonic integrated circuits (PICs), in which a variety of semiconductor optoelectronic devices are monolithically integrated and interconnected with waveguides such as lasers and modulators.
Fig. 4.1. Cable television modulator. This IOC consists of two modulation sections as well as optical power splitting and recombining regions.
Applications for integrated optics have historically been in niches of the analog, digital, and sensor fiber-optic markets; at present, however, major new markets are emerging. Perhaps the largest new market is telecommunications, where IO devices will be used for multigigabit data transmission, signal splitting and loop distribution, and bidirectional communication modules. A second new market is CATV, where IO modules will be used for external modulation in fiber-optic-based signal distribution systems. In both telecommunications and CATV, IO devices enable signal transmission at higher data rates and over longer distances. In a third market, instrumentation, a major application is fiber-optic gyroscopes. High-speed telecommunications and fiber gyro applications are common to both the U.S. and Japanese markets today. In the United States, there has been high interest in CATV and also other analog fiber-optic link applications of IO technologies. In Japan, NTT's push for fiber-to-the-home (FTTH) is driving telecommunications loop applications. A recent market forecast study of IO modulators predicts a 24% annual growth rate in North America over the 1993 - 2003 period (Tamir 1987). A significant portion of this growth is predicted to be for aerospace and military applications (e.g., fiber gyros). The forecasted annual sales by 2003 is nearly $200 million. The photonics market enabled by IO modulators (e.g., transmitters and gyros) is many times larger and will exceed $1 billion.
In the United States, several companies, most notably Crystal Technology, AT&T, and Uniphase Telecommunications Products, Inc. (UTP), sell LiNbO 3 modulator devices in the commercial market. Collectively, these companies provide devices for OEM applications in CATV, high-speed telecommunications, and fiber gyros, in addition to providing various laboratory and custom products, such as linearized modulators that meet the full NTSC 80-channel CATV specification and multigigabit/second digital modulators for 1.5 micron systems. In Japan, modulators primarily for the multigigabit telecommunications market are sold commercially by Sumitomo Cement internationally and fabricated by Fujitsu, NEC, and Oki, mainly for their own uses. The specifications of these devices are comparable to or slightly better than those of U.S. devices, but Japanese companies do not offer the variety of devices made in the United States. Numerous high-quality R&D efforts are going on in both countries. Some of the most notable development results include 16 x 16 switch arrays (16 inputs, 16 outputs) and 40 GHz modulators. In the area of glass/silica passive waveguide devices, there are only two significant U.S. producers, AT&T and PIRI (although PIRI is predominately Japanese-owned). In contrast, many Japanese companies are in this business. 1 x N splitters ( N = 4, 8, 16) are offered by Hitachi Cable, Furukawa, Fujikura, NEL, and NSG. The same companies are developing planar lightwave circuits (PLCs) for two-wavelength home terminals and bidirectional communication. Closely coupled to this work are efforts to pigtail (attach) multiple fibers at once to an IO circuit. R&D efforts in both Japan and the United States are focused on large PLCs, including planar erbium-doped amplifiers, components for wavelength division multiplexing (WDM), and structures for the silicon microbench. Much of the leading research is performed at AT&T in the United States and NTT in Japan. In the semiconductor modulator area, the major focus is on developing devices suitable for 10 Gbit/s communications. Hitachi is working on a discrete modulator, and an integrated diode laser and electroabsorption modulator is under development at AT&T in the United States and at NTT, NEC, and Fujitsu in Japan. In both the United States and Japan there is also ongoing development in the area of polymer modulators. These devices are viewed by some researchers as the next-generation commercial device, but significant development issues remain. At present, there is some research activity in Japan, most notably at NTT, and there is commercial activity in the United States at Akzo/Nobel, a part of the European parent company, which offers thermo-optic (low-speed) modulators and switches.
Multifunction modulator chips for fiber-optic gyros
Fiber-optic gyros (FOGs) are generally envisioned as the next-generation gyros for various aerospace and commercial applications (see Chapter 6). FOGs with IOCs provide the best dynamic range and sensitivity and offer the prospect of low manufacturing cost. The pioneering work and the best-performing aerospace FOGs and FOG integrated optical circuits have been developed in the United States under government sponsorship. FOG IOCs are formed in LiNbO 3 and are multifunctional because they provide beam-splitting, polarization, and modulation. These chips and the attached fiber pigtails are adequately robust for the aerospace environment and enable performance for FOGs in the intermediate grade (1 - 10 deg/hr) and inertial grade (0.1 - 0.01 deg/hr). Figure 4.2 shows a schematic of an IOC-based FOG.
Fig. 4.2. Modular FOG design.
In the United States, companies such as Honeywell and Litton are the leaders in aerospace FOG development and also have some in-house IOC capability. Commercial IOCs (based on annealed proton-exchange waveguides in LiNbO 3) are available and widely used. In Japan, JAE and Mitsubishi Precision are also developing aerospace FOGs; there is limited availability of FOG IOCs manufactured in Japan.
For nonaerospace use of FOGs, Hitachi Cable has the leading effort, although U.S. firms are also investigating these markets. Hitachi Cable is particularly focused on developing low-cost FOGs and has an effort to make FOG IO chips as an alternative to purchasing all its chips. The IO chips are needed for the company's commercial gyros, which are in the intermediate accuracy range, and also for very high-volume production of low-accuracy (1 deg/sec) gyros needed for automotive navigation systems. (These systems combine gyroscopes and GPS data with displays to form a map display in an automobile that indicates the vehicle's location and progress.) Hitachi is developing automated FOG assembly equipment and is investigating the automation of pigtailing fibers to IOCs to meet the aggressive cost targets of the automotive industry.
Active and passive IOCs for fiber-optic telecommunications
IOCs are being used in telecommunications for high-speed modulation, signal splitting and switching, and bidirectional communication. Figure 4.3 illustrates the variety of uses of modulators in telecommunications.
Fig. 4.3. IO modulator use in telecommunications. In the area of multigigabit long-haul signal transmission, there is considerable interest in external modulation as a means of minimizing fiber dispersion and laser chirp problems that limit transmission distance and signal bit rate. LiNbO 3 modulators are being used in 2.5 Gbit/s (OC-48) systems to enable transmission over distances of greater than 100 km without repeaters. Typical performance characteristics of these devices include < 5 v drive voltage and 4 Db insertion loss. Similar devices are the technology of choice for present and future undersea systems (e.g., transoceanic links operating at 5 Gbit/s), because they have negligible (or tunable) chirp and enable the greatest distance between optical repeaters, which are expensive. Modulators for 10 Gbit/s (OC-192) transmission are just now being investigated by telecommunications equipment suppliers for their next-generation systems.
External modulation is also well suited for the WDM systems now under development in the United States especially. LiNbO 3 modulators make it possible to use CW (continuous-wave) 1.5 micron lasers that have closely spaced wavelengths for transmission over the standard fiber already installed, which is optimized for 1.3 micron operation. Since installing new fiber is a major cost, the externally modulated multigigabit approach is a significant cost-saver for long-haul telecommunications operating companies. With such systems, 10 Gbit/s system operation is achieved by multiplexing four wavelength channels at 2.5 Gbit/s each. In Japan, the focus is almost exclusively on achieving 10 Gbit/s directly, without WDM, in part because NTT is installing dispersion-shifted fiber (optimized for 1.5 Ám), and the link spans are generally shorter than those in the United States.
To utilize LiNbO 3 modulators effectively in commercial telecommunications systems, manufacturers have focused on two technological issues: minimizing bias drift and achieving Bellcore-type certification on the packaged modulator. For the former issue, several approaches have been successfully developed: to passively bias the device so DC voltage is not required; to provide a special chip coating to eliminate charge buildup; and to refine the crystal growth to minimize impurities and precisely control crystal composition (stoichiometry). Bellcore certification requires devices to meet a series of environmental tests; the manufacturers have made improvements in the optomechanical packaging technology and in the humidity resistance of the fiber/chip pigtail joints to satisfy these requirements. Modulators are presently being used to transmit commercial telecommunications traffic.
As an alternative technology to LiNbO 3 external modulation, AT&T and a number of Japanese companies are developing distributed feedback (DFB) laser diodes with monolithically integrated electroabsorption modulators. These devices are intended for 2.5 Gbit/s and 10 Gbit/s systems at 1.5 Ám. Excellent laboratory results have been obtained. Relative to transmitters with CW diode lasers and LiNbO 3 modulators, the integrated structures offer lower drive voltage (2 V vs. 5 V) and a more compact size, similar to DFB lasers; however, they also have lower extinction ratios (< 20 db), residual chirp, and are proving difficult to manufacture. (See Chapter 5 for more details.)
In both Japan and the United States there are a number of excellent R&D activities in LiNbO 3 IOCs that should lead to future commercial system enhancements. AT&T has developed a polarization scrambler that significantly enhances (by several decibels in signal-to-noise ratio) the performance of undersea systems that have many optical amplifiers. Bellcore and UTP have developed acousto-optically tunable filters and switches that enable switching of closely spaced (at 2 nm) wavelength channels in WDM systems at several-microsec rates. NTT and Hewlett-Packard have demonstrated modulators with electrical bandwidths in excess of 40 GHz. AT&T, Oki, and NEC are developing high-speed switch arrays (polarization independent, complexity 4 x 4 to 16 x 16) that can be reconfigured in submicrosecond times and pass multigigabit data streams. Oki is developing a wavelength-conversion device for selectable wavelength sources in the 1.5 micron region.
IOCs formed in glass or silica films are often referred to as planar lightwave circuits (PLCs). PLCs are now available commercially for 1 x N splitters and are being developed for dual-wavelength fiber-to-the-curb (FTTC) terminals and for dense-wavelength demultiplexing. Emphasis is on low cost and environmental reliability. For splitters where N > 4, there are strong arguments for integrated structures with simultaneous pigtailing of an N-fiber array. PLC devices are being formed with ion-exchange waveguides in bulk glass by Nippon Sheet Glass (NSG) and in silica thin films by PIRI, Hitachi Cable, Furukawa, NTT, and others. For these devices, propagation losses are quite low (0.1 dB/cm), and circuits are formed on 3-inch and 4-inch substrates.
For the future, researchers in the United States and Japan envision and have demonstrated complex silica PLCs that will enable demultiplexing closely spaced multichannel WDM signals and on-chip, erbium-doped waveguide amplifiers. Also envisioned are polarization-independent thermo-optic switches. Figure 4.4 illustrates a chip under development at both NTT and AT&T that spatially demultiplexes closely spaced wavelength channels.
Fig. 4.4. Arrayed-waveguide grating demultiplexer.
A closely related activity still at the R&D stage at AT&T and NTT is silicon microbench work. Taking advantage of silicon's ability to be precision-etched, researchers are forming micromechanical structures to support and align other photonic components, such as diode lasers and LiNbO 3 modulators, with silica waveguides or optical fibers. This is projected to be a cost-effective method of interconnecting dissimilar photonic components for FTTC and FTTH applications.
IOC modulators for analog radio frequency applications
LiNbO 3 modulators are being deployed, primarily in the United States, in a significant portion of the fiber-optic CATV signal distribution systems, which utilize analog radio-frequency (RF) transmission. These systems are often of a tree/branch configuration, as illustrated in Figure 4.5. In the supertrunk portion of the network, the trend is to 1.5 microns externally modulated 80-channel systems that utilize CW laser diodes, LiNbO 3 modulators, and fiber amplifiers. This permits spans of over 50 km with no electrical regeneration. IO modulators are used in transmitters for both headend and hub locations.
For the distribution part of the network (i.e., nodal points where the signal is sent out to N lines), LiNbO 3 analog modulators in conjunction with a high-power solid-state laser can replace four or more directly modulated DFB diode lasers.
Fig. 4.5. Typical CATV system.
In order for modulators to work effectively in CATV, it is necessary to improve the linearity of conventional structures. Two different approaches have been commercialized. One is to build an electronic predistortion circuit that compensates for the modulator electrical-to-optical transfer characteristic; the other is to build a special integrated optical structure with two modulators that are driven differently to cancel nonlinear terms. These needs of the CATV industry have driven IOC manufacturers to build devices with highly reproducible RF characteristics (such as very flat frequency response up to 1 GHz).
Modulators are also very important for other fiber-optic analog applications such as antenna remoting. Over the frequency range of 100 MHz to 18 GHz, modulators in conjunction with high-power lasers have provided the highest-performance fiber links for military and aerospace applications, that is, those in the 100 dB/Hz dynamic range. Systems are now being introduced for cellular and wireless applications that utilize external modulation.
University research and other R&D activities
Osaka University and the University of Tokyo have long-standing high-quality research efforts in LiNbO 3, and efforts are ongoing in semiconductor structures. Work is at the basic materials and processes characterization level and also in demonstrating novel devices. Interest in LiNbO 3 modulators for RF link applications is centered in Japan at Osaka University and several national labs. Osaka University has also developed hybrid integrated optical disk pickup heads and quasi-phase-matching structures for diode laser frequency doubling.
In the United States, university work on guided wave devices is centered at the Universities of California at San Diego, Florida, and Wisconsin, and at Stanford University; their activities are focused on LiNbO 3 in, respectively, linearized modulators, high-speed and high-optical-power modulators, modulators with gain, and frequency-doubled structures. The University of Florida is also pursuing glass guides, and numerous universities are working on semiconductor guided wave structures, often as part of laser diode programs.
In Japan, interest in quasi-phase-matching in LiNbO 3 for frequency conversion is being researched at Oki for 1.5 micron applications and at Sony for frequency doubling (blue light generation). Sharp is not pursuing frequency doubling for blue light/data storage applications, because of the considerably larger size of such a source compared to a diode laser.
Relative to high-speed telecommunications, R&D efforts in PICs and integrated DFB lasers with electroabsorption modulators are going on at least at Oki, Fujitsu, NTT, and Hitachi. These integrated devices are appealing because of their low drive voltage and package size similar to that of regular diode lasers, but there are nontrivial development and manufacturing issues regarding chirp, fabrication yield, and emission wavelength control.
There appears to be relatively little work on polymer waveguides and modulators in Japan. While there have been several literature reports, efforts have been confined to laboratories at universities and at NTT-Ibaraki, among other places.
Another related area of work is microlens arrays. Microlens arrays are two-dimensional sets of lenslets: the state of the art is 300,000 lenslets on a 4 x 5 in. glass plate with near-diffraction-limit focusing. The lenslets are generally formed by ion exchange using techniques similar to those for forming ion-exchange waveguides. The pioneering work was done at the Tokyo Institute of Technology, and major activities are at Hoya and NSG. These lenslet arrays are designed for laser printing; another application under research is two-dimensional optical signal processing.