The intrinsic strength and fatigue of oxide glasses

Recent studies on the strength of glass fibers suggest that the time is ripe for new, fundamental studies in this area which may significantly advance our understanding of the intrinsic strength of glasses. In order to set the stage, in this paper we define various terms (intrinsic and extrinsic strength and inert and environmental fatigue) and analyze techniques for their measurement. We illustrate and evaluate these parameters by means of literature data on silica and E-glass. In addition we present some preliminary new data on E-glass fibers using 2-point bending. These new data report higher strength than previously reported and some possible reasons for this are given. While these comments deal primarily with the science of strength, a better understanding of these issues may lead to improvements in glass technology and glass products.     

Stress corrosion of silicate glass: a review

In the first section of the present paper, some examples, from the field, of the manifestations or consequences of the fatigue of silicate glass are briefly presented. In the second section, the interpretation of fatigue in terms of stress corrosion is reviewed: the role of ambient molecular water is well established. Whatever the details of the mechanism of action of water, it takes place very efficiently in the highly strained material close to the tip of a surface crack. This enables its sub-critical growth to be explained. But it does not explain many other observed effects of the environment on the mechanical behavior of silicate glass, and, in the last section, questions and issues are presented, which would still need to be investigated.     

Anomalous temperature dependence of sub-critical crack growth in silica glass

The effects of temperature, water vapor, and stress on the rate of sub-critical crack growth (SCG) in fused silica are reported. The crack velocity was measured using the double-cleavage-drilled compression method. In contrast to other inorganic oxide glasses, crack growth velocities (in region I) were found to decrease with increase in temperature. Hence a small temperature rise has the apparent effect of improving the mechanical strength of a stressed-glass part. Despite the anomalous temperature dependence, SCG in fused silica is still likely governed by the established water-enhanced stress-corrosion mechanism; another competing phenomenon is proposed to cause the observed temperature dependence. Measured crack velocities are described using an empirical model (for region I) and a mass-transport model limited by Knudsen diffusion (for region II).     

Crack path instabilities in DCDC experiments in the low speed regime

We studied the low speed fracture regime (10⁻⁴-10⁻⁹ m s⁻¹) in different glassy materials (soda-lime glass, glass ceramics) with variable but controlled length scale of heterogeneity. The chosen mechanical system enabled us to work in pure mode I (tensile) and at a fixed load on double cleavage drilled compression specimen. The internal residual stresses of studied samples were carefully relaxed by appropriate thermal treatment. By means of optical and atomic force microscopy techniques fracture surfaces have been examined. We have shown for the first time that the crack front line underwent an out-of-plane oscillating behavior as a result of a reproducible sequence of instabilities. The wavelength of such a phenomenon is in the micrometer range and its amplitude in the nanometer range. These features were observed for different glassy materials providing that a typical length scale characterizing internal heterogeneities was lower than a threshold limit estimated to few nanometers. This effect is the first clear experimental evidence of crack path instabilities in the low speed regime in a uniaxial loading experiment. This phenomenon has been interpreted by referring to the stability criterion for a straight crack propagation as presented by Adda-Bedia et al. [Phys. Rev. Lett. 76 (1996) 1497].     

Mechanical reliability of silica optical fibers

Dynamic and static mechanical tests were implemented using a tensile test bench and a static fatigue test under uniform curve. The incidence of aging treatments at 65 and 85 °C was investigated on two standard silica optical fibers (with polyacrylate and fluorinated coatings). Microscopic observations helped the understanding of the failure mechanism. It appears that the cyclic variations of the failure stress phenomenon, with respect to the aging time, are the result of the silicate gel which migrates towards the polymer coating.     

Infrared fibers based on Te-As-Se glass system with low optical losses

Unclad optical fibers based on high-purity Te-As-Se glasses prepared by chemical and physical methods of purification have been drawn. The optical, thermal and mechanical properties of glasses and fibers were investigated. The minimum optical losses were 0.07 dB/m at 7.3 μm for Te₂₅As₄₀Se₃₅ glass fiber and 0.04 dB/m at 6.7 μm for Te₂₀As₃₀Se₅₀ glass fiber. Sixty five percent of input power of a tunable CO₂ laser emitting at 9.3 μm was transmitted through a 1 m long fiber with diameter of 900 μm.     

Critical parameters in the processing of engineered stress profile glasses

A new approach to chemical strengthening is to move the maximum residual compression below the glass surface. With appropriate processing, this approach leads to strengthening and a decrease in the strength variability. This improvement in mechanical reliability is accomplished by introducing stable crack growth into the glass, which leads to multiple cracking of the glass surface prior to failure. In the current study, a fracture mechanics analysis is presented that uses a simple form for the stress profile. By determining the effect of this stress profile on the apparent fracture toughness of the material, it is possible to identify the conditions when spontaneous cracking occurs, the stress for the onset of multiple cracking and the ultimate strength of the glass. In particular, it is shown there is an important interplay between the depth of the maximum compression, the stress gradient and the magnitude of the maximum compressive stress. The current analysis is then used to define a strategy for the processing of glasses with engineered stress profiles.     

Crack growth threshold in soda lime silicate glass: role of hold-time

Atomic force microscopy was used to examine the shapes of cracks and residual features left behind on the fracture surface after holding cracks at a stress-intensity factor below the fatigue threshold for soda lime silicate glass. After propagating a crack at a stress-intensity factor of K_I=0.37 MPa m^1/2, the crack was held either at K_I=0.24 MPa m^1/2, or K_I=0.1 MPa m^1/2 for periods ranging from 1 h to approximately 200 h. Cracks held at the higher K_I left featureless lines on the fracture surface. Becoming more pronounced as the hold-time increased, these lines marked the position of a corrosive notch that formed during the hold period. At about 30 nm from the crack tip, the crack surface displacement decreased reaching zero at the crack tip. The crack tip shape was that of an ogee arch. At the lower hold valued of K_I, crack bifurcation occurred during which the crack became wavy, part of the crack propagating into the crack plane, part out of the crack plane when the crack was repropagated. A smaller crack tip displacement was observed for these cracks. Results of this study are believed to be a consequence of corrosion of the fracture surface caused by a basic solution formed when alkali ions (Na⁺ and K⁺) at the crack tip exchange with hydrogen ions in solution.     

Application of mercury porosimetry to the study of xerogels used as stone consolidants

Alkoxysilanes, low-viscosity monomers that polymerize into the porous network of stone by a sol-gel process, are widely used in the restoration of stone buildings. We have used the mercury porosimetry technique to characterize changes in microstructure of three granites following their consolidation with two popular commercial products (Wacker OH and Tegovakon V). The suitability of this technique is questioned because a surprising increase of stone porosity is observed. In order to investigate the feasibility of porosimetry, we analyze the behavior of xerogels prepared from the two commercial products, under mercury pressure. Gels are basically compacted and not intruded by mercury. Thus, the increase of stone porosity after consolidation can actually be associated with gel shrinkage. Mercury porosimetry, therefore, has been found unsuitable for characterizing the microstructure of consolidated rocks. However, it can be employed usefully to evaluate shrinkage of gels under mercury pressure, which permits the behavior of a consolidant during the process of drying in stone to be predicted. It is a key factor because many problems of consolidants are related to their drying process within the stone. Gels under study exhibit a high rigidity and an elastic behavior, as consequence of their microporous structure. Finally, the reduction in the porous volume of gels after the porosimetry test demonstrates that the shrinkage mechanism is based on pore collapse.     

The thermodynamics of water and hydrogen solubility in fused silica

A statistical thermodynamic model of the solubility of water vapor in glass is developed as an extension of previous models of monatomic and diatomic gas solubility in glass. The model predicts the $\text{p}{}^{\mathrm{<mtext>1</mtext><mtext>2</mtext>}}$ dependence of water vapor solubility in fused silica and indicates the binding energies of dissociated H and OH species to be about ?63 and ?99 kcal/mol, respectively. The OH binding energy is found to be slightly temperature dependent, namely E_OH(0) = ?81.3?0.0163T. The tendency of chemical solubility of H₂ to predominate over physical solubility above 500°C correlates with the model equations for H₂ solubility.     

Sol-gel chemistry approach in the preparation of precursors for the substituted superconducting oxides

Sol-gel processing of ceramic materials for advanced applications involves several steps starting from precursor synthesis and ending up with multicomponent metal oxides. A simple sol-gel synthesis technique has been refined to prepare the precursors for the superconducting (Y_1−xSc_x)Ba₂Cu₄O₈ and (Y_1−xGa_x)Ba₂Cu₄O₈ compounds. The amorphous gel powders were characterized by powder X-ray diffraction, infrared spectroscopy, thermal analysis and elemental analysis. A systematic characterization of precursor gels led us to predict the approximate composition and the chemical reactions involved during gelation. The stability and high level of homogeneity obtained for the gels make them suitable as processable precursors to substituted (Y_1−xSc_x)Ba₂Cu₄O₈ and (Y_1−xGa_x)Ba₂Cu₄O₈ superconducting compounds.     

The effect of implanting nitrogen on the optical absorption and electron paramagnetic resonance spectra of silica

Silica samples (type III, Corning 7940) were implanted with N using multiple energies to produce a layer ∼600 nm thick in which the concentration of N was constant to within ±5%. The optical absorption spectra of the samples were measured from 1.8 to 6.5 eV. Electron paramagnetic resonance (EPR) measurements were made at ∼20.3 and 33 GHz for sample temperatures ranging from 77 K to 100 K for most measurements. The components identified in the EPR spectra, based on comparison with reported parameters, were due to E′ centers and peroxy radicals. By comparing the changes in the optical absorption at 5.85 eV with the changes in the concentrations of the various EPR components and with the reports in the literature, we conclude that there is an additional band at 5.7-5.9 eV other than the E′ center band. We conclude that the bands between 2 and 6.5 eV and the EPR spectral components produced by implantation of N are due to radiation damage processes; neither optical bands nor EPR components related to N are detected.     

Micro- and mesoporous silicas synthesized in acidic water-ethanol solution of equimolar catanionic surfactant

Porous silicas with combined micro- and mesoporosity are synthesized in acidic water-ethanol solution of equimolar catanionic mixture, where the mesopores are narrowly and uniformly distributed, and the micropores generate due to the addition of ethanol. To vary the pH value of the synthetic mixture can also change the ratio of micro-/mesopores volume in the resulting samples. Compared with other amorphous silica gels and ordered mesoporous silicas including MCM-41, MCM-48 and SBA-15, these micro- and mesoporous silicas show much improved adsorptive capacity for volatile nitrosamines.     

Photoluminescence in as-drawn and irradiated silica optical fibers: an assessment of the role of non-bridging oxygen defect centers

The spectral and temporal characteristics of the photoluminescence in as-drawn and irradiated silica and doped silica fiber-optic waveguides have been investigated. The extended pathlength available with a fiber-optic geometry has offered the opportunity to make high sensitivity emission measurements on high silica glasses under both steady state and pulsed laser excitation. The analyses of the fiber data coupled with emission studies on selectivity doped bulk glasses suggest that the dominant emission band centered near 650 nm is intrinsic to defects in the SiO network, specifically, dangling non-bridging oxygens ions which can be generated by irradiation, fiber drawing or by the introduction of network modifying ions such as alkali.     

The role of the main silica precursor and the additive in the preparation of low-density xerogels

The incorporation of an additive during sol-gel synthesis reduces shinkage during ambient drying. The following additives have been studied: 3-(2-aminoethylamino)propyltrimethoxysilane (EDAS), 3-aminopropyltriethoxysilane (AES) and 3-(2-aminoethylamino)propyltriethoxysilane (EDAES) and the main silica precursors were tetraethylorthosilicate (TEOS) and tetrapropylorthosilicate (TPOS). When the additive contains methoxy groups (EDAS), it acts as a nucleation agent of the silica particles and exactly the same properties (pore volume, specific surface area, particle and aggregate size) are obtained whether the main reagent is TEOS or TPOS. The nucleation mechanism is based on the difference in reactivity between additive and main reagent. In case of nucleation by the additive, the nucleation agent fixes the properties whatever the main silica precursor is. When both the additive and the main reagent contain ethoxy groups (series AES-TEOS and EDAES-TEOS), there is no nucleation mechanism by the additive, and the silica particle size remains nearly constant. With less reactive main reagent (series AES-TPOS and EDAES-TPOS), pore volumes up to 17 cm³/g have been obtained with pore sizes up to nearly 10 μm and very big particles (∼100 nm). The absence of nucleation by the additive for the couples AES-TPOS and EDAES-TPOS could be due to the fact that the difference in reactivity between ethoxy groups and propoxy groups is not sufficient to initiate the nucleation mechanism by the additive. In the absence of nucleation by the additive, the main reagent plays a role: highly porous materials with very large pores are prepared with TPOS.     

Silica gel with continuous macropores prepared from water glass in the presence of poly(acrylic acid)

Monolithic silica gel with macropores was prepared in the solution of water glass and polyacrylic acid (HPAA) by freezing transitional structures of phase separation. In the system, phase separation proceeds between silica polymers and HPAA, so that the porous morphology varies from closed macropores to particle aggregates through bicontinuous morphology, where both macropores and silica skeleton are three-dimensionally interconnected, with increasing HPAA/silica ratio. In addition, we can control the macropore size in bicontinuous morphology by varying the concentration of both silica and HPAA, or by changing the molecular weight of HPAA. The pore size distribution is quite sharp indicating the presence of pores with the same size all through the monolithic samples.     

X-ray irradiation-induced coloration of manganese in soda-lime silicate glass

Coloration of Mn in soda-lime silicate glass by X-ray irradiation was investigated as a coloration method suitable for glass recycling. Mn was initially reduced to Mn²⁺ with the addition of Fe, Sn or carbon, and the resultant colorless or near-colorless glasses containing no Mn³⁺ were studied. These glasses turned violet after X-ray irradiation and heat treatment at 200 °C due to formation of Mn³⁺. The color was stable at room temperature and decoloration occurred with heat treatment at 600 °C. The addition of Fe or Sn increased the optical absorption by Mn³⁺ ions. ESR measurement suggests that Mn photo-oxidation occurred through charge transfer from photo-induced hole centers.     

Improvement of the Tm:H4 level lifetime in silica optical fibers by lowering the local phonon energy

The role of some glass network modifiers on the quantum efficiency of the near-infrared fluorescence from the ³H₄ level of Tm³⁺ ion in silica-based doped fibers is studied. Modifications of the core composition affect the spectroscopic properties of Tm³⁺ ion. Adding 17.4 mol% of AlO_3/2 to the core glass caused an increase of the ³H₄ level lifetime up to 50 μs, 3.6 times higher than in pure silica glass. The quantum efficiency was increased from 2% to approximately 8%. On the opposite, 8 mol% of PO_5/2 in the core glass made the lifetime decrease down to 9 μs. These changes of Tm³⁺ optical properties are assigned to the change of the local phonon energy to which they are submitted by modifiers located in the vicinity of the doping sites. Some qualitative predictions of the maximum achievable quantum efficiency are possible using a simple microscopic model to calculate the non-radiative de-excitation rates.     

Free energy profiles of Al and La cation distribution in silica and soda silicate glasses

The factors that control the distribution of Al³⁺ and La³⁺ cations in silica and soda silicate glasses is examined by using molecular dynamics (MD) simulations. In particular, the response of the glass network to the presence of metal oxide is probed using a liquid state theory that treats the glass network as a solvent and the metal cation as a solute. MD simulations are used to obtain the mean solvent-solute and solute-solute force. The trajectory used to determine the free energy is analyzed to determine the stable configurations of a cation pair. Details of determining the potential of mean force for an Al cation pair in silica and silicate glass is presented. A comparison of these results with those previously calculated for a La cation pair in the same glass systems is given. The results reveal that there are differences in how the network accommodates the two different size cations. It is found that for the potential used here, based on the Vessal potential, the network wraps itself around the larger La cation forming a solvent shell, whereas, the smaller Al cation is incorporated into the network backbone. In silica and soda silicate glasses, La ion pairs cluster to form La-O-La linkages. In contrast, the glasses favor a separated state of the Al ion pair.     

The effect of HF/NH4F etching on the morphology of surface fractures on fused silica

The effects of HF/NH₄F, wet chemical etching on the morphology of individual surface fractures (indentations, scratches) and of an ensemble of surface fractures (ground surfaces) on fused silica glass has been characterized. For the individual surface fractures, a series of static or dynamic (sliding) Vickers and Brinnell indenters were used to create radial, lateral, Hertzian cone and trailing indentation fractures on a set of polished fused silica substrates which were subsequently etched. After short etch times, the visibility of both surface and subsurface cracks is significantly enhanced when observed by optical microscopy. This is attributed to the increased width of the cracks following etching, allowing for greater optical scatter at the fracture interface. The removal of material during etching was found to be isotropic except in areas where the etchant has difficulty penetrating or in areas that exhibit significant plastic deformation/densification. Isolated fractures continue to etch, but will never be completely removed since the bottom and top of the crack both etch at the same rate. The etching behavior of ensembles of closely spaced cracks, such as those produced during grinding, has also been characterized. This was done using a second set of fused silica samples that were ground using either fixed or loose abrasives. The resulting samples were etched and both the etch rate and the morphology of the surfaces were monitored as a function of time. Etching results in the formation of a series of open cracks or cusps, each corresponding to the individual fractures originally on the surface of the substrate. During extended etching, the individual cusps coalesce with one another, providing a means of reducing the depth of subsurface damage and the peak-to-valley roughness. In addition, the material removal rate of the ground surfaces was found to scale with the surface area of the cracks as a function of etch time. The initial removal rate for the ground surface was typically 3.5× the bulk etch rate. The evolving morphology of ground surfaces during etching was simulated using an isotropic finite difference model. This model illustrates the importance that the initial distributions of fracture sizes and spatial locations have on the evolution of roughness and the rate at which material is removed during the etching process. The etching of ground surfaces can be used during optical fabrication to convert subsurface damage into surface roughness thereby reducing the time required to produce polished surfaces that are free of subsurface damage.