In weakest link models the failure of a single microscopic element of a brittle material causes the failure of an entire macroscopic specimen, just as a chain fails if one link fails. Pristine samples of glass, such as optical communications fiber, approach their ideal strength, and their brittle tensile failure has been descried by this model. The statistics of weakest link models are calculable in terms of the statistics of the individual links, which, unfortunately, are poorly known. Use of the skewness of the failure distribution may permit simultaneous determination of the statistics of the individual weak links and of their number density, which indicates their physical origin. However, the applicability of weakest link models to real materials remains unproven.

Using Two-Point Bending and Tensile tests in commercial optical fibers it was found that strength results are connected. Because of that connection many fractures cannot be explained assuming isolated defects. Other explanations are proposed in terms of a family of defects spread in length and superficial stress areas originated during manufacture.

In this paper we present the results of computational studies on a set of experimental data concerning the strength of commercially available silica optical fibers. We compare various statistical methods for characterizing the distribution of fiber strength and assess their validity by comparing random sampling from a set of experimental data with results generated using a Monte Carlo computer routine.

Strength measurements with a broad spectrum (from less than 2 GPa to more than 5 GPa), at different stress-rates have been analyzed. The number of samples for each stress-rate was large (200). For each stress-rate, two straight lines, one for high and the other for low fracture strength can fit the strength data in a Weibull diagram. Intersection of both lines depends strongly on stress-rate. In order to obtain the strength and fatigue behavior for several strengths, the data for each stress-rate were divided in six group of data points by considering different ranges of probability (i.e., inert strength). Good linear fittings are obtained in Weibull diagrams for most groups. The fatigue parameter n is larger for the lowest and highest strength groups (n equals 22) and decreases for the middle groups (lowest n equals 15). In the lowest strength group, a large deviation from the linear fitting is observed, mainly for the lowest strength points. This behavior suggests the operation of some random mechanism.

Residual stress in an optical fiber impacts several fiber properties, including reliability and geometry. The residual stress profile arises from a thermal expansion mismatch of the constituent materials, the tension applied during fiber forming, and the thermal profile experienced by the fiber during formation. The thermal profile of the fiber is determined by measuring the intensities of the two `defect' bands in the silica cladding of the fiber using Raman spectroscopy. Other studies have shown that the intensities of these bands increase with increasing fictive temperature and that these changes can be used to map changes in fictive temperatures across the fiber diameter. Optical birefringence is used to measure the residual stress profile of the fiber. The relationship between fictive temperature and fiber residual stress is explored for both straight fibers and fibers with a natural radius of curvature (curled fibers).

The reliability of optical fiber exposed to relatively high static strains (> 2%) has been extensively modelled and investigated by experiment. Fatigue `knees' have been demonstrated predicting the premature fracture of fiber particularly where elevated temperatures and relatively large volumes of water have been used to soak the samples. The cause has been attributed to simultaneous stress- assisted and stress-free corrosion of the fiber surface. In this paper we show that, a t more moderate strains (1 to 2%) and using a limited volume of water, there is evidence of a strength recovery caused either by a healing process or the observance of some form of lower strain threshold. The expected strength reduction of the fiber, from contemporary models is contrasted to that observed. The unusually high strength retention shown by the test fiber in water is shown to have important implications for optical cable design and for the bending of fiber within joint housings.

Polymeric coatings are the most important strength reliability factor of fiber optics. Diverse accelerated aging processes have been used in the last years with the objective to estimate the lifetime of fibers in real field situations. Because the humidity present at the glass- polymer interface determines the strength, a dry technique in fiber optics is used to understand the properties of coating permeability after aging process. Dry technique parameters can be connected with the strength evolution after months in hot water aging. To compare the efficiency of several coatings under this technique different fibers from different manufacturers were used. The strength measurements were made using the Two-Point Bending apparatus.

Optical fibers have been found to exhibit an accelerated rate of strength reduction during static fatigue and zero stress aging for long times in aggressive environments. This phenomenon has been commonly referred to as the fatigue and aging `knee'. The onset of the knee has been found to be highly variable and is sensitive to the polymer buffer coating. In past work we have shown that moisture vapor penetrates most polymer coatings on the time scale of minutes, which implies that the diffusion rate of small molecules is not the rate-determining step for aging. On the other hand, the diffusion of large molecules through the polymer coatings can take anywhere from weeks to years to reach the polymer/glass interface. The implication of this result is that large molecule diffusion might be the rate- determining step in aging. In the work presented here the diffusion of moisture and pH buffer solutions through various optical fiber coatings will be discussed. These results are correlated with the zero stress aging behavior of the same fibers.

Static fatigue measurements have been carried out on silica fibers for telecommunication networks. These fibers were immersed into water at temperatures ranging from 20 degree(s)C to 70 degree(s)C. They were winded around alumina materials, which induced a permanent stress lying between 3.2 and 3.5 GPa. The evolution of failure time with respect to temperature is Arrhenian. We propose a general relation with semi empirical constants, from which the failure time may be calculated as a function of temperature and stress. The n factor and parameters were calculated. The n factor decreases linearly versus temperature while the BS_i^n-2 parameter increases exponentially with respect to temperature. This reflects two different phenomena: the structural relaxation which compensates the external stress and the acceleration of the chemical attack of the water molecules.

Since the classic paper of A. A. Griffith in 1920, questions have been asked about the variability of the strength of glass fibers as a function of composition and processing conditions. While Griffith apparently found that the strength of soda-lime-silica fibers increased as their diameter decreased, he also found that quite large-diameter silica fibers had strengths of the order of 6.9 GPa (1 X 10⁶ psi). Since that time, almost all investigators have found very similar strength values for silica regardless of the fiber diameter, raw material or conditions of formation. The history of the strength of fused silica fibers and the effect of various coatings on the strength, fatigue and aging are reviewed in this paper.

Short summary of the achievements and conclusions of the European research action COST 246, titled Materials and Reliability of Passive Optical Components and Fiber Amplifiers in Telecommunications Networks, which was current during 1993 - 1998, is given.

Subcritical crack growth in fused silica can be modeled as a stress assisted chemical reaction between water and strained bonds at the crack tip. The stress influences the crack growth rate by reducing the free energy of the activated complex. In principal, the stress changes both the activation enthalpy (energy) and entropy; however, the influence of stress on entropy has generally been ignored. The dynamic fatigue behavior of `pristine' optical fiber can be used to determine the fatigue kinetics parameters with unprecedented precision. It is shown that the entropy contribution is at least as significant as the enthalpy and therefore should not be ignored.

It was obtained that (t₂S_i²)-((sigma) _s/S_i) and ((sigma) _d/Si₁)-((sigma) '/S_i³) are universal coordinates for presentation of static fatigue and dynamic fatigue data respectively. Usage of these coordinates helps to correctly compare the results of tests of different kinds of fibers (strong and weak) regardless of the initial defect size. presentation of the dynamic fatigue data for pristine and indented fibers in universal coordinates showed a very similar behavior of both fiber types in spite of their difference in strength.

A single-layer UV-curable polyacrylate-coated telecommunications grade fused silica fiber was found to have a significant reduction in two-point bending strength after immersion in acetone. The two-point bending and tensile strengths of this fiber as a function of immersion time in acetone were determined, and this strength loss was not seen for 0.5-m gauge length tensile specimens. SEM and optical fractography was performed on the weak specimens, and the cause of the strength reduction is proposed to arise from particles smaller than 3 micrometers in the coating. These particles could cause surface flaws by sliding contact damage incurred during relative motion between the coating and the glass. This sliding could occur either while flexing the fiber in preparation for a bending strength measurement or due to coating elongation. While it is not clear which mechanism is operating, both are consistent with the observation that degradation is only observed for bending strength measurements.

The mechanical reliability of optical fiber used in certain biomedical applications is extremely important because failure of the fiber during use might be fatal for the patient. Therefore, prediction of the lifetime of the fiber both in storage and during service is necessary before the fiber can be safely used. In this paper we study two commercially available optical fibers designed specifically for high power laser delivery. The fatigue parameters calculated from static fatigue data are used to estimate the maximum allowed stress that ensures survival for the deign life of the fiber. This work properly accounts for uncertainty in the predictions; uncertainty which arises not only from scatter in the experimental data, but also from uncertainty in the form of kinetics model to use for extrapolation (i.e. power law, exponential, etc.). This paper thus provides an outline for making lifetime predictions for a critical applications involving relatively short lengths of fiber, that does not bind in any questionable assumptions.

Fictive temperatures of an optical fiber surface were measured using IR reflection spectroscopy with a microscope attachment. The measurement was possible because the IR peak wavenumber of the silica structural band exhibited a good correlation with the fictive temperature of silica glasses. The fictive temperature of the silica optical fiber surface was extremely high, higher than that of the interior of the fiber by several hundred degree(s)C. On the other hand, during the fiber drawing, the temperature difference between the surface and interior of the fiber is estimated to be small, approximately 10 degree(s)C, and the difference in cooling rate between the fiber surface and interior is considered negligible. The observed high fictive temperature of the fiber surface was attributed to the high tensile stress of the surface layer during the fiber drawing.

It is a general assumption that water itself, rather than impurities of water solutions, is dangerous for standard optical fibers. In this review paper, a summary of the results on water test conditions, made by COST 246 Action, is given, and the factors affecting the test results of strength and fatigue tests, are discussed. A fiber may have very different strength degradation in water depending on the chemical conditions, such as, ion concentration of the water, test vessel material, temperature etc.

Atomic force microscopy has proven to be of great value in the study of the surface roughness of aged silica fibers. It has been shown that aging in both liquid water and water vapor results in surface roughening which correlates with and is apparently responsible for the strength degradation. In this paper, we show that the application of a new indenting/scratching/imaging tip and associated software allow this tool to be extended to the study of a new range of problems. We illustrate the usefulness of this `nanoindentation/imaging probe' in the study of indents and scratches on silica fiber surfaces which have undergone a variety of treatments as well as in study of coatings on these silica lightguide surfaces.

Photopolymerization of a number of neat acrylate monomers used in polymer coatings of optical fiber was studied with photoDSC and with a cure monitor using a fluorescent probe. Acrylates had functionality from one to six. PhotoDSC results show that conversion of monomers ranges from 40 to 100% depending upon functionality and structure of a monomer. Kinetics of hardening of a sample under light at ambient temperature was nicely fit into two-exponential law; rate constants k₁ and k₂ in this empirical analysis are in the range 0.5 - 35 min^-1. Dependence of rate of polymerization and conversion upon functionality of a monomer is discussed. Both cure monitor and photoDSC can be successfully used for testing of liquid coatings. It was demonstrated, that an application of a moderate permanent magnetic field increases rate of cure of coating up till 20%.

The mathematical and numerical analysis of two nonlinear problems of solid mechanics related to the breaking strength of coated optical glass fibers are presented. Both of these problems are concerned with the two-point bending technique which measures the strength of optical fibers by straining them in a bending mode between two parallel plates. The plates are squeezed together until the fiber fractures. The process gives a measurement of fiber strength. The present theory of this test is based on the elastica theory of an unshearable and inextensible rod. However, within the limits of the elastics theory the tensile and shear stresses cannot be determined. In this paper we study the behavior of optical glass fiber on the base of a geometrically exact nonlinear Cosserat theory in which a rod can suffer flexure, extension, and shear. We adopt the specific nonlinear stress-strain relations in silica and titania-doped silica glass fibers and show that it does not yield essential changes in the results as compared with the results for the linear stress-strain relations. We obtain the governing equations of the motion of the fiber in the two-point bending test taking into account the friction between the test fiber and the rigid plates. We develop the computational methods to solve the initial and equilibrium free-boundary nonlinear planar problems. We derive formulas for tensile and shear stresses which allow us to calculate tension in the fiber. The numerical results show that frictional forces play an important role. The interaction of optical fiber and rigid plates is treated by means of the classical contact theory.

The aim of this representation is to show that by using near field optical microscopes it is possible to obtain results about the optical cladding surface of optical fibers. We used two near field microscopes: a Shear-Force Microscope and the Photon Scanning Tunneling Microscope. We will discuss the effect of the aging and of the process of stripping on the microstructures observed on the two kinds of images: topographical and optical.

Fiber Bragg gratings are used in a wide variety of devices including sensors, tunable filters, and signal controllers for Wavelength Division Multiplexing. Bragg gratings can be formed in an optical fiber by illuminating the fiber from the side with a pattern of ultraviolet light. Most gratings are made using 240-nm light. However, by using 330-nm light the grating can be written right through the standard polymer coating of the fiber, which preserves the fiber's mechanical strength. We discuss some of the mechanisms that degrade the strength of fiber gratings. We also discuss applications of mechanically strong fiber gratings, including very wide (> 50 nm) tunable filters.

Among several possible failure modes in a Fiber Bragg Grating (FBG) device, fracture of the optical fiber is one of great importance. Reliability of fiber in such a device has to be ascertained and assured. A technique to evaluate the fiber strength in this device has been developed, and the extent of degradation due to processing and handling has been established. The impact of mechanical failures in other parts of the device on fiber failure is also addressed and evaluated. The failure mechanisms and its implications on fiber reliability are discussed. A proof stress level has been determined and implemented in the fabrication to assure mechanical reliability of the fiber against time. Based on the fiber strength distribution, proof stress level used, and the applied stress, a FIT rate is calculated using power law crack growth model for silica fibers. This study estimates an average FIT of 0.06 at ambient room temperature over a 25 year life for fiber failure in FBG devices fabricated by Corning Inc.

The focus of this paper are performance and mechanical and optical reliability of fiber Bragg gratings as stress/strain and temperature sensors in high temperature applications for extended periods of time. However, not a particular sensor application will be considered but functionality and reliability of fiber Bragg gratings under this condition will be investigated.

This study presents results from an experiment where Bragg gratings were used to measure the stress on a bare optical fiber placed inside a 3M VF-45 connector. The connector exerts lateral compressive and longitudinal tensile stresses on the fiber, both of which yield a shift in the Bragg wavelength. The axial stress is of interest for mechanical reliability predictions, so additional measurements were performed in order to separate the two effects. This involved simulating the use of an alternate material in the connector to induce differential longitudinal stress on the fiber grating.

We present results on optical and mechanical stability of single-layered acrylate coated fiber Bragg gratings produced on a draw tower, and exposed to high temperature annealing and to cyclic loading. Optical stability was assessed in terms of strain response and reflectivity changes with annealing temperatures up to 400 degree(s)C. Cyclic loading of Bragg gratings with mean stresses between 2.4 GPa and 3.0 GPa was compared with predictions made by using power-law based crack growth theory with parameters obtained by dynamic tensile tests. Comparison with theory confirms predicted strength decrease and lifetime reduction.

We present a measurement setup for the complete characterization of fiber Bragg gratings and wavelength selective fiber-optic devices. We can measure the spectral response of these devices both in transmission and in reflection, the wavelength dependency of the group delays due to chromatic and Polarization Mode Dispersion (PMD), as well as the wavelength dependency of the Polarization Dependent Loss (PDL). Experimental results are presented and sources of error are discussed. Comparisons with the Jones matrix method for the measurement of PDL and group delay due to PMD are made.

The optical attenuation induced in 6 Navy Qualified Parts List fibers by ⁶⁰Co (gamma) -ray exposures to 10⁶ rad has been measured at 3 different dose rates: 26, 1000, and 20,500 rad/min. The loss in the Corning, Lucent and Plasma Optical Fiber multimode fibers was measured at 1340 nm, while that in the Corning, Lucent, and SpecTran single mode fibers was measured at both 1315 and 1510 nm. Growth and recovery data were measured for the highest does rate exposure, while only growth data were obtained for the lower dose rates. In an effort to predict the fibers' radiation response at much lower dose rates, the data have been fit to a model based on an exponential distribution of trap depths below the glass mobility edge. Although the results of the model are in qualitative agreement with the data, additional refinement is necessary. However, the agreement is sufficient that an upper bound can be calculated for the degradation induced by low-dose-rate radiation environments.

The results of a 5000 hour life test at 85 degree(s)C/85% RH and of a 5000 hour life test at 45 degree(s)C/85% RH on commercial planar and fused-fiber optical branching devices are presented. The failure mode of planar optical branching devices is revealed by an increase of excess loss with a parallel increase of back reflection whereas the failure mode of fused-fiber optical branching devices is evidenced by splitting ratio variation. Failure analysis suggests possible degradation mechanisms for both types of devices. Furthermore Mean Time To Failure range has been estimated. Experimental results show that highly reliable couplers could be manufactured provided that a short period screening procedure is performed.

Erbium-doped silica optical fiber amplifiers have emerged as an effective means of restoring the power of optical signals due to losses caused by attenuation, interconnect components or signal splitting. With these optical fiber amplifiers, modest lengths of the amplifier fiber can be placed at appropriate distances along the optical fiber network to boost the signal as needed. Although short lengths, typically less than ten meters, of fiber are required, the mechanical strength and reliability of Er-doped fibers are still essential for their successful use in the telecommunication industry. This paper reports the results of mechanical strength characterization of four-meter gauge lengths of single mode Er-doped amplifiers. Dynamic fatigue strength measurements reveal that these fibers have nearly unimodal strength distributions with median strengths in the range of 4.8 to 5 G Pa and Weibull slopes in the range of 16.6 to 69. Measured N-parameters of 24.4 +/- 0.7 to 27.4 +/- 0.6 for these fibers are higher than expected for acrylated urethane coated fibers and are comparable to N- parameters measured for standard telecom fiber. That the measured N-parameter before cycling (or 24.4 +/- 0.7) does not differ significantly from the N-parameter after 85C/85% RH-cycling (or 24.4 +/- 0.6) suggests very good resistance of these fibers to environmentally adverse conditions.

In this paper we describe a single kinetic model consistent with current experimental results of the effect of hydrogen on erbium doped fiber, and derive three approximations to the model that can be used to fit data from fiber. Fits to data are shown for two of the three models.

The optical attenuation induced in two erbium-doped fibers by ⁶⁰Co (gamma) -ray exposures to 200 krad has been measured at 3 different dose rates, approximately, 35, 1000, and 20,500 rad/min. The growth of the induced loss was measured at 980, 1300, and 1500 nm. In some instances recovery of the loss after exposure was also measured. The effect of temperature on radiation-induced loss was examined by irradiating one of the erbium-doped fibers at three different temperatures, -54, 30, and 80 C. A simple physical model describing the thermal annealing process of the induced loss in the fibers is presented. The model yields expressions, with only two adjustable parameters, which describe the dose rate and temperature dependence of both the radiation-induced loss growth and recovery kinetics. The simple model accurately describes the form of the growth and recovery kinetics but overestimates the dependence of induced loss on dose rate and temperature.