Search

Cooper–Krohn analysis of stress generation in soda lime silicate (“SLS”) glasses during ion exchange strengthening suggests that the glass should develop 2.5 to 3 GPa surface compression (biaxial stress), though only ~ 600 MPa values are observed experimentally. In this paper, we suggest that this “network dilation anomaly” is largely resolved. MD simulations were carried out to monitor volume and pressure changes after ion-swap. Surface compression in an SLS glass ion-exchanged as a function of temperature and time in KNO 3 bath was measured to follow the viscous relaxation. We show that biaxial surface compression builds initially to the expected ~ 2.5 GPa; thereafter fast secondary structural (“β”) relaxations, whose time constants span picoseconds to nanoseconds, cause the surface compression to relax rapidly to ~ 1.5 GPa at the ion exchange temperature. Stress relaxes thereafter due to “α” primary structural relaxations associated with viscous flow. Recovery occurs to a transition state which is somewhat short of the expected glass volume. Full volume is recovered at higher temperatures due to increased flow.

Chemical strengthening of glass by ion exchange below its glass transition temperature has become a well-established technological process in the glass industry since its introduction in the 1960s. However, some questions related to the basic science of the ion exchange process still remain. In this paper we focus on the observation of a maximum in compression that is often observed below the surface, rather than at the surface as predicted by the Cooper theory. In some glasses, this anomaly appears in a dramatic form, where the surface compression reduces to zero and can even become tensile upon continued ion exchange. We provide an explanation of this anomaly based on a mechanism of volume shrinkage (densification) triggered by the large hydrostatic pressure inherent to ion exchange in some glasses. We propose a mechanical spring-and-dashpot model for the pseudo-viscoelastic behavior of glass to identify the mechanisms and associated time scales that can reproduce the experimental observations. Next, we modify Sane and Cooper's analysis by introducing a new “ V -factor” to account for the identified relaxation effects (fast β-relaxation, slow α-relaxation, and stress release due to volume and shear viscosities). The approach accurately reproduces the subsurface compression maximum anomaly. Our model therefore enables a clearer perspective on the evolution of the glass network during the chemical strengthening process.

23Li2O·77SiO2 glass with nanodimensions was grown within the pores of a pellet formed by ZnO nanorods of a diameter of 70 nm and lengths varying from 80 nm to 270 nm. The resistivity of the nanoglass was less than that of its bulk counterpart by more than three orders of magnitude. The activation energy was found to be 35% lower than that of the bulk glass. This was ascribed to an increase in free volume in the nanodimensional glass. The ac conductivity variation at high frequency showed an exponent of 0.5 indicating two dimensional movements of lithium ions in the nanoglass. Nanocomposites of lithia–silica nanoglass and CuO nanoparticles showed magnetodielectric property with dielectric permittivity at 500 kHz decreasing by 5.3% at 2 T. This was explained on the basis of space charge polarization and Hall Effect. Nanocomposites of lithia–silica nanoglass and ZnO nanorods exhibited a giant magnetodielectric effect with the dielectric permittivity increasing by 20% to 99% at 2 T depending on the measuring frequency. This was also ascribed to space charge polarization with the ZnO nanophase showing a negative magnetoresistance.

The vibrational behavior of binary silicate glasses is modeled using molecular dynamics (MD) simulations. The prevailing procedure for assigning vibrational modes to structural features is through comparison with spectra of known substances. The explicit knowledge of atomic trajectories from MD simulations allows for a direct observation of vibrational motion in specific frequency bands and decomposition into the predominant vibrational modes. We demonstrate a Fourier transform method for determining the vibrational motion associated with a particular frequency. We compare literature assignments for α-cristobalite, vitreous silica, β Na2O 2SiO2 and vitreous Na2O 2SiO2 to this method. Our analysis shows that while it is necessary for the network to be disrupted and non-bridging oxygen atoms to form for vibrational motion to occur at 1205 cm− 1, the mode is in fact not specific to that structural moiety.

Chemically strengthened glasses are of increasing technological importance for personal electronic devices, tablet computers, and a variety of other applications. However, there are many unexplained phenomena associated with the physics of the ion exchange process used for strengthening. One of the most puzzling of these is the anomalous behavior of the network dilation coefficient, i.e., the parameter governing the amount of linear strain of the glass per unit of alkali ions exchanged, which is inevitably a factor of 2–4 higher for as-melted glasses as compared to chemically strengthened versions of the same glass compositions prepared via ion exchange. In this paper, we investigate the atomistic origin of this discrepancy between as-melted and ion-exchanged glasses based on molecular dynamics simulations of a series of alkali tetrasilicate glasses, viz., x Na 2 O·(20 − x )K 2 O·80SiO 2 (mol%). The network dilation coefficient of the ion-exchanged glasses is dependent on composition and ranges from 30% to 54% of the ideal value obtained from the as-melted glasses. This anomalous behavior of the network dilation coefficient is explained in terms of different local environments between sodium and potassium sites in the glass network and a two-stage relaxation process of the local potassium environment after ion exchange.

Sodium tracer diffusion coefficients, D⁎Na, have been measured in sodium borosilicate glasses of the type (Na2O)0.2(B2O3)y(SiO2)0.8−y as a function of temperature and the composition parameter y. At constant temperature, the tracer diffusion coefficient of sodium decreases as y increases. The activation enthalpy derived from sodium tracer diffusion data for temperatures up to about 350 °C increases about linearly with increasing values of y from about 70 to 100 kJ/mol. The pre-exponential factor of the sodium tracer diffusion coefficient as a function of y varies by about one order of magnitude and has a minimum at near y = 0.3.

Glasses strengthen during an ion exchange experiment as a result of the high surface compression that develops from the stuffing of the invading alkali ion into the smaller host alkali ion site and the non-relaxation of this compression. The kinetics are described in terms of a Nernst–Planck interdiffusion coefficient of the exchanging ions in a stack of mixed-alkali glass layers with varying ratios of the alkali. The interdiffusion coefficient is suppressed by compression and enhanced by tension. However, the measured surface compression is usually 2–4 times lower than that calculated using thermal stress analogy incorporating an elastic suppression of the molar volume difference of the as-melted mixed-alkali glasses. A new view of the physics of deformations and failure of stress to continue to buildup is presented in an effort to seek technology improvements. Reduced stresses are ascribed to network yield in two separate modes: yielding of the shape-conserving hydrostatic stress component which prevents elastic expansion of molar volume and yielding of the volume-conserving deviatoric (pure shear) stress which leads to measurable shape change. An anomalous subsurface tension is also explained in this view. By drawing analogy to microindentation, it is concluded that optimum network topology glasses should undergo the least plastic deformation and, hence, develop a higher surface compression.

We propose a generalized definition of entropy accounting for the continuous breakdown of ergodicity at the laboratory glass transition. Our approach is applicable through all regimes of glass forming, from the equilibrium liquid state through the glass transition range and into the glassy state at low temperatures. The continuous loss of ergodicity during the laboratory glass transition is accompanied by a loss of entropy as the system gradually becomes trapped in a subset of the configurational phase space. Using a hierarchical master equation approach, we compute the configurational entropy of selenium, a simple but realistic glass-former, for cooling rates covering 25 orders of magnitude, viz., 10 −12 to 10 12 K/s. In all cases, the entropy of glass is zero in the limit of absolute zero temperature, since here the system is necessarily confined to a single microstate.

We describe the potential energy landscapes of Se3–Se8 clusters using disconnectivity graphs. Inherent structures include both chain and ring configurations, with rings dominating in the larger clusters. Transition points are located using an eigenvector-following technique, which is of general applicability in mapping multi-dimensional energy landscapes.

We develop model interaction potentials for the binary As–Se system using ab initio molecular simulations and a cluster expansion technique. These potentials are used with classical Monte Carlo simulations to characterize the structure of As 2 Se 3 , AsSe 2 , and AsSe glasses. Finally, we compute the fraction of soft modes in the As–Se system as a function of average coordination number. The results show evidence of a rigidity percolation threshold at an average coordination number between 2.3 and 2.4.

Elastic/plastic properties of Ge–Se binary and Ge–Sb–Se ternary chalcogenide glasses were explored by indenting with Vickers and Brinell (spherical) microindenters using static and ‘instrumented’ (load versus penetration) recording machines and analyzed further using a non-contact profilometer. The load-dependence of microhardness, indentation fracture toughness and brittleness versus the average covalent coordination number, 〈r〉, were studied. It is concluded that the elastic recovery of deformations is maximized at ∼GeSe4 (〈r〉 = 2.4), which is consistent with ‘optimized connectivity’ of the rigidity percolation arguments. Surprisingly, many of the extrema were nearly non-existent in the ternary glasses.

We derive a new model of the glass transition for isobaric conditions. Our model is based on mapping the continuous (3N + 1)-dimensional enthalpy landscape of a glassforming system to a discrete set of inherent structure and transition point enthalpies. Using a master equation approach, we follow the time evolution of a system from its equilibrium liquid state through an arbitrary cooling path. Our model employs a phase space distribution matrix to enable computation of macroscopic properties as a function of time. We have implemented our model for selenium and present the first-ever volume–temperature curves of a glassforming system based solely on ab initio physics.

This paper is a brief review of our previously published results in addition to a recent study of the microhardness of chalcogenide glasses measured as a function of the average covalent coordination number, 〈 r 〉. Glasses in the pseudoternary system Ge–(Sb·As)–(Se·Te) were prepared by vacuum melting of previously distilled 5N to 6N pure raw materials. Several physical properties, namely, density (molar volume), elastic moduli, static and dynamic microhardness, indentation toughness, thermal expansion coefficient, glass transition temperature, viscosity, and acoustic attenuation have been measured and published previously. Most glassy state properties have a monotonic relation with 〈 r 〉. Molar volume and the configurational contribution (difference between liquid state and glassy state properties) to second derivative thermodynamic properties, on the other hand, have minima at the rigidity percolation threshold 〈 r 〉=2.40. Newly obtained microhardness data using a recording indenter show that binary chalcogenide glasses have a dynamic Vickers microhardness, called `DVH' achieving a maximum at 〈 r 〉=2.4. Whereas the energy required for plastic and unrecovered elastic deformations is largest for Se (〈 r 〉=2.0 glass), the elastic energy recovery upon unloading the indenter appears to be a maximum at 〈 r 〉=2.4. Based on these observations I suggest that the 〈 r 〉=2.4 structure has the largest yield strength and, in turn, is the most connected network among the binary glasses.

A mixed-alkali aluminosilicate glass (glass transition temperature, Tg=460∘C) was ion exchanged at 375∘C in a fused silica cell in the hot-stage of a polarizing microscope. Buildup of stress-induced optical birefringence was measured as a function of the depth below the surface and time of exchange to 8 h. Surface compression increased steadily during this period. Upon rapidly cooling the specimen to room temperature, stresses decreased ∼20%, without a significant change in the thickness of the compression layer implying that the surface layer had a higher thermal contraction following the exchange. Differential thermal contraction of ion-exchanged layers could not have been the cause of either large differences between calculated stress magnitudes from those observed experimentally, or the observed “anomalous” tensile stress maximum. It is argued that the ion-exchanged layers are stuffed derivatives having physical properties different from those of the parent glass or the as-melted glasses of corresponding compositions.

To understand the effects of structural features and to locate their signatures in the glassy and the liquid states, various properties were studied as a function of average coordination number, 〈 r 〉, in chalcogenide glasses and examined in relation to the chemically ordered covalent network model and constraint theory. The former predicts a threshold at 〈 r 〉 = 2.67 and the latter a threshold at 〈 r 〉 = 2.4. It is noted that most of the properties exhibit a threshold at 〈 r 〉 = 2.67 while only some show the threshold at 〈 r 〉 = 2.4. The reasons for the existence of the thresholds and various practical applications based on composition-property relationships are discussed.

High purity chalcogenide glasses with an average coordination number, 〈r〉, between 2 and 2.8 were prepared by vacuum melting of pre-distilled samples in the GeSe binary and GeSbSe ternary systems. Viscometry and scanning calorimetry were performed to understand the nature of the property thresholds observed with respect to 〈r〉 in these glass-forming systems. There is a lack of correlation between viscosity-based and heat capacity-based classifications of strong and fragile liquids in the overconstrained region of these glass-forming systems. An argument based on Adam-Gibbs equation and mode decoupling is put forward for the observation of the anomaly.

High purity chalcogenide glasses with an average covalent coordination number, 〈 r 〉, in the GeSe binary and GeSbSe ternary systems between 2 and 2.8 were prepared by vacuum melting pre-distilled elements. To understand the effects of 〈 r 〉 on glass-forming capability, properties such as thermal expansion coefficient, molar volume and heat capacity were studied as a function of 〈 r 〉. Prior authors have searched, generally fruitlessly, for extremum behavior either in glassy state properties or in liquid state properties of glass-forming compositions at 〈 r 〉 = 2.4 to support Phillips' constraint theory. The missing link is provided if one examines the configurational changes during glass transition at ordinary cooling rates. The configurational contributions to the heat capacity and thermal expansion, in addition to the molar volume, show distinct minima at 〈 r 〉 = 2.4, suggesting that the structural rearrangements for the 〈 r 〉 = 2.4 liquid in the glass transition region are minimized. If such a liquid possesses minimized accessible structural rearrangements in the supercooled liquid region as well, then it may be concluded that such a liquid would display a poor crystallization tendency. This, then, renders support to Phillips' argument that 〈 r 〉 = 2.4 solid with degrees of freedom equal to the number of constraints marks the ‘best’ glass.

Stress-induced optical retardation at λ = 1.8 μm in GeAsSe chalcogenide glass disks subjected to a diametral compressive load was measured using a modified IR-polarizing microscope. The computed stress-optic coefficient, B, varied from −29.6 to +5.4 TPa−1 over the glass-forming region. Although the magnitude of B appeared to decrease with increasing Ge content, there appeared to be little direct correlation between B and the average coordination number, 〈(ifr)〉.

High purity mid IR-transparent chalcogenide glasses are generally prepared by vacuum melting of 5–6 N pure raw materials such as Ge, Sb, Se, As and Te. For laboratory samples, a modified technique is proposed which involves a combination of surface oxide removal prior to, and distillation, melting and annealing after, batching the raw materials without breaking vacuum. Se, As and Te were distilled in sequence simply by increasing the temperature in steps. The need for porous quartz frit used by previous researchers during distillation was eliminated, thus reducing the possibility of impurity inclusions.

Several high purity chalcogenide glasses were prepared by vacuum melting of previously distilled and/or surface oxide removed 5 to 6-nines pure Ge, Sb, Se, As and Te. Vickers hardness number and indentation toughness were measured using the microindentation technique. A previously proposed empirical relation between average coordination number and Vickers hardness number for chalcogen-rich chalcogenide glasses was refined and is discussed in relation to ‘rigidity percolation’ concepts.

High purity chalcogenide glasses incorporating Ge, Sb, Se, As and Te were prepared by vacuum melting of previously distilled 5 N to 6 N pure raw materials from which the surface oxide was also removed in some cases. Molar volume and elastic properties of several isostructural chalcogenide glasses prepared by similar processing techniques were obtained from the measured values of densities and the velocities of transverse and longitudinal acoustic waves respectively. Property variations with the average coordination number, 〈r〉, were examined. Molar volume displayed a distinct minimum at 〈r〉 = 2.4, coincident with Phillips' ‘percolation threshold’ for each system studied. Elastic moduli, however, failed to show any dramatic changes at this threshold. Near the stoichiometric tie-line, a chemically ordered covalent network (COCN) model for the atomic arrangement in these glasses was found to be preferred over the chance coordination predicted by the random covalent network (RCN) model.

High purity chalcogenide glasses incorporating Ge, Sb, Se, As and Te were prepared by vacuum melting of previously distilled 5 N to 6 N pure raw materials from which the surface oxide was also removed in some instances. Glass transition temperatures, Tg, of several chalcogenide glasses prepared by similar processing techniques were determined using differential scanning calorimetry (DSC). An emperical relationship between the glass transition temperature and average coordination number is proposed by modifying the Gibbs-DiMarzio equation for glass transition of a cross-linked polymer as a function of cross-link density. The results for the Tg of three ‘iso-structural’ systems are reported and discussed in light of presumed structural arrangements in these glasses.

Glasses covering the entire glass formation region in the GeSbSe chalcogenide system were prepared using 5- to 6-nines pure raw materials and careful processing techniques. Vickers microhardness and surface toughness of these, as well as several other isostructural glasses prepared similarly, were made using the microindentation technique. The microhardness of most of the glasses increased linearly with the average coordination number 〈 r 〉. The surface toughness showed a maximum at 〈r〉 ⋍ 2.6 . The empirical relationships between glass composition and the two properties so derived are examined, to some extent, in light of percolation concepts in these covalently bonded solids.

Shelyubskii wavelength scan technique to measure glass homogeneity is described. The Raman-Shelyubskii theory, which is based upon interference optics and which, hitherto, was believed to be the underlying mechanism, is discarded. A computer simulation of light scattering based on geometric optics principles fits the observed transmission vs wavelength curves very well.

DC ionic conductivity was measured over 300–800 C in specimens of type-I fused silica doped with various levels of Al2O3/M2O ratio (M = alkali) using blocking and non-blocking electrodes. Although the values of the conductivity depended upon the measurement technique, all the data showed an Arrhenius behavior with the values of activation energies ranging from 35 kcal/mole to 20 kcal/mole for various alkali ions below 500 °C. A slight decrease in activation energy was seen in most of the samples at temperatures greater than 500 °C. When the resistivity at a particular temperature was plotted as a function of Al2O3/M2O ratio, a broad minimum in resistivity was seen between the ratios of 4–12. These results are discussed in terms of the AlO4− and nonbridging oxygen sites in the fused silica structure.

Potential methods for decreasing the weight of glass containers are reviewed. The greatest potential lies in methods for producing high surface compressive stresses. Glazing, controlled crystallisation and ion-exchange techniques appear most promising, but all have serious disadvantages.

The structure of amorphous AlPO 4 was calculated using the method of molecular dynamics. Calculated radial distribution functions are in good agreement with those measured by Wignall et al. using X-ray diffraction. The structure, although similar to that of SiO 2 in many features, lacks full connectivity and does not appear as random. This may help explain the resistance of AlPO 4 to glass formation.