It is encouraging to see a meeting of experts gathering at this early point in the life of fiber optic systems to consider in technical depth the control of reliability. In the past reliability has often been a consideration left to relatively late stages of the development of a new technology. In the heat of the battle to achieve any kind of functionality for a new technology, it is hard to think ahead to achieving long, reliable 'lifetimes, but we must constantly remind ourselves to begin to solve this problem at the earliest point possible.
The long term reliability of high silica glass fiber lightguides used in lightwave communications systems is a critical consideration for assuring necessary system performance over design lifetimes. Three important concerns have emerged: the long-term mechanical strength of glass fiber; radiation induced changes in optical attenuation; and hydrogen effects on optical loss. This paper will highlight the key aspects of such phenomena, with particular emphasis on current understanding of silica fiber lightguides used for telecommunications applications.
We have investigated effects of polymer coatings on the strength and fatigue properties of silica fibers. In this study, we performed tensile strength, dynamic fatigue, and static fatigue measurements on silica fibers coated with a variety of uv curable polymers. We have identified the coating properties that influence the mechanical reliability of optical fiber. Polymer coatings can provide adequate protection against mechanical abrasion and microbending loss. However, they cannot protect fibers from fatigue because most polymers are highly permeable to water vapor. Among fibers coated with polymers that have different water permeation and adhesion properties, there are considerable variations in fiber strength and fatigue behavior. We observed that both adhesion and water absorption properties of the coating material affect fiber strength and time-to-failure. We showed that the static fatigue parameter can be significantly lower than the dynamic fatigue parameter when the fiber coating has a low water-absorption value. A lifetime estimate of optical fiber based on the dynamic fatigue parameter is unreliable without consideration of the coating properties; and a more reliable estimate should be based on static fatigue testing. The selection of polymer coatings with high adhesion-to-glass and low water absorption values seems to benefit the mechanical reliability of optical fiber. In summary, coating is an integral part of optical fiber in any mechanical reliability consideration of a polymer-coated optical fiber. Adhesion and water permeation properties of a polymer coating will affect the mechanical reliability of optical fiber.
Mechanical failure of optical fibers under stress over extended period of time is caused by static fatigue of fibers. To ensure long term reliability, a method of mini-mizing the static fatigue phenomenon in optical fibers is necessary. Potential solution is to put an impervious hermetic coating around the optical fiber so that diffusion of moisture to glass can be arrested. A technique for putting such a hermetic coating on optical fiber is chemical vapor deposition. Performance results of optical fibers coated with several different materials are presented.
Mid infrared laser technology centered around 10pm has been driven by the availability of highly efficient, high power tunable CO2 lasers, continuously tunable lead salt diode lasers, and the rapidly advancing infrared fiber technology. Utilization of such systems is already becoming commonplace in medicine and laser heterodyne spectroscopic systems for remote sensing. These applications are increasingly requiring system flexibility, mobility, hence compaction and integration. To meet these needs, some steps toward 10 pm system integration and compaction have been explored with available mid IR components. We will report on our studies in the CO, laser + metallic piping/fiber + detector system and the tunable diode laser (TDL) + fiber detector system for heterodyning. Metallic piping will be compatible with w/g CO2 laser and w/g packaged detectors, where the 1.f. beat will be in the microwave region. Fibers will be compatible with TDL for direct butting, but incompatible with w/g packaged detectors. There is also severe thermal gradient between fiber section in the TDL coldhead at LH2 or LN2 temperatures, and section at room temperature. Current technology provides lowloss, rugged, near single mode piping, but appreciably higher loss, fragile, chemically unstable multimode fibers. However, there is no doubt fibers will be the ultimate mid IR laser system integration medium, so most of our efforts are in fiber testing and fiber system integration. Comparison studies have been done on relevent fiber parameters, such as loss, toxicity, hygroscopicity, refractive index, flexibility, shelf life, and thermal behavior at low temperatures of crystalline KRS-S, halide and chalcogenide glass fibers. Emphasis is placed on thermal shock evaluation at LN2 and LH2 temperatures with respect to mode structure, flexibility and shelf life.
Activities of standards groups such as parameter definition and product and test method specification can influence component and system reliability. In addition to these tradi-tional activities, lightwave standards groups are beginning to take a more direct role in product reliability by creating procedures for quality conformance specification, process control and monitoring, and the qualification approval of manufacturers. Perhaps the best known function of standards groups is illustrated by their famous agreement in the late 1970's on 50/125 pm as the optical fiber core/cladding dimensions for trunk applications. There were several fiber sizes being offered at the time,1 and the agreement probably contributed to the rapid early growth of the lightwave industry. Standards groups engage in other activities besides product specification, and these activ-ities can affect product reliability in both direct and indirect ways. Some of the more common activities are summarized in this paper.
Semiconductor injection lasers are key components in fiber optic communication systems, therefore their reliability is a determining factor in the technical success of these systems. Lasers are subject to several specific degradation mechanisms some of which affect the bulk of the optically active volume of the device, some the mirror facets and some the electrical or thermal contacts. These mechanisms are driven primarily by the operating temperature of the device or by the locally produced electrical and optical power density, but their severity is strongly linked to technological factors such as the quality of the crystal growth (substrate and epitaxial layers) and subsequent processing technologies (mirror passivation, mounting, con-tacting, etc.). They also depend on the chosen materials (GaAs/A1GaAs or InP/InGaAsP), that is the emission wavelength, contact materials and mirror passivating materials. The best way to assure high reliability of these devices in the field - in addition to perfecting and controling the manufacturing processes - is the development of accelerated aging tests and the selection of strict and meaningful criteria for the acceptance or rejection of a particular device. Such tests and criteria have been developed supported by experiments and theoretical predictions. It can be shown that carefully manufactured lasers do not degrade more rapidly than LED-s and their mean time to failure at normal operating conditions may even exceed those of other active components of the system such as driver electronics, etc.
A more generalized (Eyring) type of acceleration model for laser diodes is discussed which includes temperature, bulk current and optical power as accelerating stresses. This may allow greater accuracy in laser diode lifetime estimation and in prediction of the lifetime implications of changes in laser diode structural design. The effect of redundancy on laser diode lifetime requirements, diode parameters essential to fiber optic and free-space communication system operation, screening procedures, failure analysis, test configurations and environmental/equipment safeguards are also discussed.
Packaging is a key element in attaining high reliability in fiber-optic sources and detectors. In many regards, the challenges of developing stable packaging equal or exceed the challenges faced in manufacturing the critical semiconductors elements. For example, in the development of semiconductor elements it is possible to use high temperature lifetests of a few months to allow prediction of room temperature lifetimes of many tens of years. For packaging, such an extrapolation is not straightforward and may not be possible. However, the conventional wisdom is that there is a significant increase in package degradation rates at elevated temperatures. Consequently, it is generally accepted that packaging design efforts should be aimed at techniques which demonstrate good stability at high temperatures.
A highly reliable PINFET has been developed and manufactured by all solder construction using passivated InGaAs PIN photodetector. At operation voltage of -5V, the PIN pnotodetector has typical characteristics of 3nA in dark current, 0.3pF in capacitance and 0.75A/W in responsivity. The PINFET receiver circuit is designed using these PIN photodetectors. Extensive SPICE simulation has been performed to reduce the effect of stray capacitances thus to improve the manufacturing yield and quality to meet the 3σstatistical limits for all the parameters. Lifetests have been performed on the 16 PIN photodetectors and 10 fully packaged PINFETs with equivalent room temperature operation of 1.7 x 107 and 2.2 x 105 hours respectively without failure.
With the data rates of LANs starting to exceed 200 Mbits/sec and long-haul communication systems entering the gigahertz realm, system designers are having to look for alternatives to standard ECL logic to achieve these ultra-high speeds. Until a few years ago, only a precious few OEMs large enough to own a GaAs fab line would have considered it practical to baseline GaAs logic as a solution. However, with the availability of off-the-shelf GaAs digital components, any manufacturer that is accustomed to dealing with ECL technology can now realize system data rates greater than 500 Mbps with ease. This paper will discuss the considerations related to the utilization of GaAs ICs in fiber optic systems including reliability, thermal management, and interconnection issues.
The performance of silicon (Si) and gallium arsenide (GaAs) active components operating in a radiation environment is discussed. These components include diode detectors, Si MOSFETS and GaAs MESFETS. The radiation environments considered include total dose gamma, high dose rate gamma, high energy heavy particles and neutrons. The total dose effect of charge trapping and interface state generation in the Silicon dioxide due to total dose gamma is discussed. The large photocurrents due either to high dose rate gamma or high energy heavy particles are discussed. Some of the process optimizations used to enhance the radiation immunity in Si technology and their relevance to a better and more reliable fiber optic system are presented.
The reliability and performance of fused taper couplers over a variety of environmental changes was investigated. Specifically, the effect of temperature on the state of polarization of light in both a twisted and an untwisted 2x2 single mode coupler operating at 632.8 nm was investigated. Also the effects of temperature and humidity on the coupling ratios of both 2x2 multimode and 2x2 single mode couplers were investigated.
Optical wavelength division multiplexers (WDM) were tested for performance degradation as a result of temperature cycling, shock and pulling on the fiber leads to determine tensile strength. The WDM components tested use multidielectric coatings to separate two channels, one ranging from 0.75 to 0.95 micron and the other from 1.2 to 1.6 microns. The optical crosstalk remained constant under the various tests and the insertion loss variation was small.
Fiber optic LAN transmission system reliability is ultimately determined by the reliability of the individual components in these systems. For LANs the optoelectronic components are predominantly GaAlAs LEDs and Si photodetectors for 830nm transmission while InGaAsP LEDs and InGaAs photodetectors are used for 1300/1500nm transmission. Component design and packaging is determined by system application and such factors as data rate, power output and fiber core size. Typical data rates for LANs vary from 2Mb - 200Mb, while fiber types of interest include single mode and 50,62.5, 85 and 100 micron multimode fibers. In this presentation we will first discuss the design and selection of reliable emitter and detector components for LANs. Secondly, a review of our reliability results for GaAlAs edge emitting LEDs and InGaAs PIN detectors will be presented. These data include five years of component reliability testing and quality assurance monitoring in a manufacturing environment.
Fiber optic technology has made a very rapid transition from the research and development stage into practical operating systems. This transition has evolved in a natural progression towards more complex and demanding technology. This progression has resulted in wide application of multimooe systems and increasing deployment of single mode technology. Finally, optical-guided-wave and integrated-optical devices are beginning to emerge into the commercial arena. In all these cases, reliability is a prime concern that must be addressed. Considerable progress has been made in this area with regard to conventional multimove and single mode components and systems. However, the reliability aspects of optical-guided-wave ((JGW) devices are just beginning to be studied. This paper will discuss reliability considerations in TiLiNb03 puided-wave devices. These considerations will include material factors, device structures and designs, and packaging and connection of optical fiber pigtails. In the first category, processing/material factors will be reviewed and photorefractive, pyroelectric, and humidity effects will be discussed. The impact of packaging and pigtailing upon reliability will be covered. Test data on pigtailed 1i:LiNb03packaged devices under theermal, shock, and vibration exposure will be presented and discussed.
Reliability is mandatory in all parts of a Fiber Optic System and especially in the interconnections - an electronically passive component yet mechanically actuated numerous times throughout the service life. The key loss mechanisms and evaluation criteria are discussed with comparisons of the butt-joint and lens type terminations for relative strengths.
This paper summarizes the state-of-the-art operational and reliability characteristics of fiber optical components typical of those used in optical transmission systems. Standby requirements for vulnerable components are evaluated by using this data in relevant reliability formulae developed for fiber optic systems having a redundant design.
Fiber optic Local Area Network Systems are being used to interconnect increasing numbers of nodes. These nodes may include office computer peripherals and terminals, PBX switches, process control equipment and sensors, automated machine tools and robots, and military telemetry and communications equipment. The extensive shared base of capital resources in each system requires that the fiber optic LAN meet stringent reliability and maintainability requirements. These requirements are met by proper system design and by suitable manufacturing and quality procedures at all levels of a vertically integrated manufacturing operation. We will describe the reliability and maintainability of Codenoll's passive star based systems. These include LAN systems compatible with Ethernet (IEEE 802.3) and MAP (IEEE 802.4), and software compatible with IBM Token Ring (IEEE 802.5). No single point of failure exists in this system architecture.
One of Bell Communications Research's (Bellcore's) functions is to perform technical analyses of fiber optic transmission systems on behalf of its clients. One segment of those technical analyses is to determine the expected steady-state reliability performance of a system and compare it to Bellcore's objectives. This paper discusses the methodology Bellcore uses to model the reliability of fiber optic transmission systems. The paper begins by discussing Bellcore's reliability objectives applicable to short haul, interoffice, fiber optic transmission systems for telephone company intra-LATA applications. The paper then discusses modeling the reliability of the electro/optic hardware modules, which is calculated using the component base failure rates, while considering the expected level of quality, burn-in, electrical stress, and probable environment. The paper next discusses techniques appropriate for modeling current system architectures and protection arrangements. These models lead to the calculation of steady state system reliability parameters such as the contribution to service unavailability due to hardware reliability failures. Finally, the results of a system reliability model calculation are shown with a comparison to the objectives.
The panel was conducted in an open forum in which all participants contributed in the discussion. The participants included the distinguished panelists and moderators listed above and the attendees of the SPIE Fiber/Lase '86 Conference in Cambridge, Massachusetts.