Access the full text.
Sign up today, get DeepDyve free for 14 days.
Glass Struct. Eng. (2020) 5:445–487 https://doi.org/10.1007/s40940-020-00138-2 SI: CHALLENGING GLASS Investigating the ﬂexural strength of recycled cast glass Telesilla Bristogianni · Faidra Oikonomopoulou · Rong Yu · Fred A. Veer · Rob Nijsse Received: 26 April 2020 / Accepted: 9 September 2020 / Published online: 2 November 2020 © The Author(s) 2020 Abstract Currently, tons of high quality commer- Young’s modulus. The prerequisites for good quality cial glass are down-cycled or landﬁlled due to con- recycled cast glass building components are identiﬁed taminants that prevent close-loop recycling. Yet, this and an outline for future research is provided. glass is potentially a valuable resource for casting robust and aesthetically unique building components. Keywords Cast glass · Glass ﬂexural strength · Exploring the potential of this idea, different types Glass defects · Recycling of glass waste · Mechanical of non-recyclable silicate glasses are kiln-cast into properties of glass 30 × 30 × 240 mm beams, at relatively low tempera- tures (820–1120 C). The defects occurring in the glass specimens due to cullet contamination and the high vis- 1 Introduction cosity of the glass melt, are documented and correlated to the casting parameters. Then, the kiln-cast speci- The great potential of glass casting technology for mens and industrially manufactured reference beams the building industry is so far little explored by struc- are tested in four-point bending, obtaining a ﬂexural tural engineers and architects, but are gradually get- strength range of 9–72 MPa. The results are analysed ting discovered after the success of all cast-glass load according to the role of the chemical composition, level bearing structures such as the Crystal Houses façade of contamination and followed casting parameters, in in Amsterdam (Oikonomopoulou et al. 2018c). The determining the ﬂexural strength, the Young’s modu- 3D-shaping possibilities provided by casting can offer lus and the prevailing strength-limiting ﬂaw. Chemi- robust glass components of larger cross-sections and a cal compositions of favourable performance are high- wider variety of forms and colours than currently avail- lighted, so as critical ﬂaws responsible for a dramatic able by other glass processing methods. Parallel to the decrease in strength, up to 75%. The defects situated in recognition of the structural and aesthetical strengths the glass bulk, however, are tolerated by the glass net- of cast glass components, questions arise regarding work and have minor impact on ﬂexural strength and their environmental impact and life-cycle. The use of - currently not recyclable-disposed glass as a raw source T. Bristogianni ( ) · R. Yu · R. Nijsse for glass casting at lower temperatures, is a promising Faculty of Civil Engineering and Geosciences, TU Delft, idea that addresses both the pressing problem of glass Delft, The Netherlands e-mail: firstname.lastname@example.org waste, and the urgency to lower the carbon footprint of glass building components (Bristogianni et al. 2018; F. Oikonomopoulou · F. A. Veer Oikonomopoulou et al. 2018b). To specify the term Faculty of Architecture and the Built Environment, TU Delft, “currently not recyclable glass”, apart from the suc- Delft, The Netherlands 123 446 T. Bristogianni et al. cessful recycling of soda lime glass food and beverage 2008; Bristogianni et al. 2019), which may compromise containers, the rest of the discarded-often high quality- the strength of the glass product. commercial glass rarely meets the strict standards of This paper explores the ﬂexural strength of recy- the manufacturers due to contamination from coatings cled cast glass—a property relevant to the engineering and/or adhesives. The lack of infrastructure for collec- practice. Aim is to give insight into the effect of the tion, product disassembly and cullet separation con- casting parameters on the strength, and to assess the cerning these different types of glass, originates from plausibility of employing waste glass for the produc- the hesitation of the manufacturers to accept this cul- tion of safe structural components. Thus, in this work, let, and thus limits or prevents its recycling. Therefore, a variety of commercial glass waste silicates is tested as this glass cannot ﬂow back into the original prod- and evaluated for their ability to be kiln-cast into struc- uct system (close-loop recycling), it gets down-cycled tural components at relatively low temperatures (820– to applications such as aggregate, ceramic-based prod- 1120 C). The occurring defects are documented and ucts, foam insulation, abrasives (Silva et al. 2017), or correlated with the stage of production during which is disposed of in landﬁlls. As the need of ﬁnding alter- they are caused. Thereafter, two series of four-point native routes, markets and end-users for the upcycling bending experiments are conducted in kiln-cast glass of the tons of high-quality discarded glass is impera- beams of 30 × 30 × 240 mm dimensions. The results tive, the partial diversion of this waste into the build- are analysed according to the role of the chemical com- ing industry by casting structural glass components is position, level of contamination and followed casting worth exploring. parameters, in determining the ﬂexural strength and the The above developments reveal a gap in the litera- origin of fracture. The testing of a limited number of ture concerning the mechanical properties of cast glass industrially manufactured components serves as a point components and the suggestion of a design strength of reference. for their structural use. This is linked with the absence of established manufacturing procedures and quality 2 Experimental work control standards, and thus the great variability in the strength of the cast glass products according to each 2.1 Glass cullet categorization and specimen manufacturer and the corresponding glass composition preparation and casting process applied. The use of waste glass cul- let is an added complication to this issue, giving rise to a This work studies a series of characteristic commercial series of traditional and new types of defects (Bartuška glasses, used for the production of common glass prod- Table 1 Specimen preparation, cullet categorization and kiln-casting settings 123 Investigating the ﬂexural strength of recycled cast glass 447 Table 1 continued 123 448 T. Bristogianni et al. ucts such as ﬂoat glass, glass ﬁbers, cookware and lab- oratory glassware, cast glass bricks, crystal ware and CRT TV screens. The choice of glasses is made in alignment with the types of waste glass cullet provided by various glass manufacturing and recycling compa- nies, in order to address the recycling of readily avail- able waste glass sources and thus tackle a realistic prob- lem. X-ray ﬂuorescent (XRF) analyses are conducted with a Panalytical Axios Max WDXRF spectrometer to deﬁne the chemical composition of the selected glasses. The provided cullet is thoroughly cleaned with iso- propanol, and alien material (metal, plastic, cork) is manually removed when possible. The identiﬁed con- Fig. 1 Arrangement of the investment moulds inside the taminants in the given cullet, still present in traces after ROHDE ELS 1000S kiln the cleaning process, are listed in Table 1 according to the following categorization: kiln (Fig. 1) and kiln-cast, meaning that only one kiln i. Coatings (soft, hard, mirror, enamel, frit) is employed for the complete casting process (heating ii. Variations in composition of the same glass type up, forming, annealing and cooling). (different manufacturer, tints) The glass samples are formed at viscosities between iii. External contaminants during sorting: a. Organ- 6 3.5 10 –10 dPa s and at top temperatures ranging from ics (e.g. plastic, textiles), b. Non-glass inorganics 820 to 1120 C that are selected according to the chem- (e.g. ceramics, stones, porcelain, glass ceramics), ical composition of each glass. The viscosity (η) range c. Metals, d. Different glass types (e.g. borosilicate, 4 1.5 chosen is higher than the 10 –10 dPa s forming and lead glass) melting range adopted by the glass industry, taking into account the risk of inhomogeneity of the ﬁnal prod- The cullet is then used for kiln-casting the 30 × uct. The approach of glass forming at lower tempera- 30 × 240 mm glass beams required for the four-point tures is chosen on the one hand to reduce the required bending tests. This particular beam size is selected as it energy and corresponding CO emissions, and on the provides a substantial thickness of cast material so that 2 other hand to intensify the occurrence of defects and the inﬂuence of the defects in the bulk can be evaluated, evaluate if their existence is acceptable for a structural while keeping the mass below 1kg, and therefore reduc- glass product. Thus for several samples (e.g. ﬂoat glass, ing the annealing time. For each glass cullet, at least borosilicate rods), 2–3 different top temperatures are 3 samples are produced for statistical purposes. The 5–6 tested, corresponding to viscosity ranges of 10 dPa s cullet is positioned inside of disposable silica/plaster 3–4 and 10 dPa s, to further study the inﬂuence of the investment moulds made from Crystalcast M248, in defects on the ﬂexural strength. All specimens are kept a structured or random manner. The moulds are then 4 ◦ at top temperature for 10 h, quenched at a − 160 C/h placed in a ROHDE ELS 200S or ELS 1000S electric Cathode Ray Tube (CRT) screen production has ceased, yet Footnote 3 continued there is still a considerable volume of CRT glass cullet resulting reported casting by melt-quenching. As relatively low forming from the separation of disposed screens (Andreola et al. 2005). temperatures are chosen, the corresponding high viscosity of In case less samples are reported, the cullet available was not the heated glass does not allow its instant pouring from a melt- sufﬁcient for the production of three samples. These specimens ing (platinum or high-alumina) crucible to a preheated (steel or are nonetheless presented in this study to demonstrate the failure graphite) mould for annealing. Thus, the whole casting process mode of the speciﬁc type of glass, rather than derive an absolute has to take place in one mould that can withstand temperatures ﬂexural strength value. up to 1150 C, does not attach to the glass and does not cause 3 fracture to the specimen during cooling. Crystalcast M248 is an investment powder consisting of 73% silica content (cristobalite, quartz), 23% calcium sulphate (gyp- Given the high viscosity at top temperature and the size of sum) and 1% organics (Goodwin Refractory Services 2003; Gold the samples, a 10 h dwell is empirically found suitable for the Star 2019). The choice of the mould material is related to the kiln- removal of large bubbles (> 1 mm) and the incorporation of the casting technique followed in this work versus the commonly coatings to the glass network. 123 Investigating the ﬂexural strength of recycled cast glass 449 rate down to their annealing point, heat-soaked for 10 h • Single 30 × 240 mm ﬂoat glass panes of a 8/10 mm and cooled down to their strain point with a − 4 C/h thickness, edges ground and polished ramp, before controllably cooled down to room temper- The grinding and polishing procedure followed for ature at a faster rate. This conservative annealing sched- the preparation of the above specimens is identical to ule guarantees stress-free specimens, as seen through the one described for the kiln-cast samples. However, cross-polarized light. the bottom and top surface of the ﬂoat glass specimens The specimens are produced at a 40 mm compo- (single and bonded) is kept in its as received condi- nent height, and then cut to size with a water-cooled tion (optically ﬁne polished) and only the cut edges are rotary diamond wheel cutter, to remove the top surface processed. that often contains a high amount of ﬂaws (e.g. surface crystallization, bubbles, depletion of alkali in the com- position, wrinkling, crazing). Then, the specimens are 2.2 Four-point bending test set up ground and polished with a Provetro ﬂat grinder and diamond abrasive discs in sequence of 60, 120, 200, 1st series of experiments (12 kiln-cast, 6 reference spec- 400 and 600 grit and their resulting dimensions are imens) documented. The inhomogeneities in the glass speci- The 1st series of experiments is conducted in order mens are observed by naked eye and with the use of to provide a general overview regarding the ﬂexural a Keyence VHX-5000 or VHX-7000 Digital Micro- behavior of the different glass specimens. The speci- scope. A qualitative assessment of the internal resid- mens are tested using a Zwick Z10 displacement con- ual stresses in the glass specimens is achieved by using trolled universal testing machine in a laboratory air crossed-polarized ﬁlters. Lastly, the beams are prepared environment and at a rate of 0.2 mm/min. The four- for the Digital Image Correlation (DIC) measurement point bending ﬁxtures have a 110 mm span for the load- by creating a speckle pattern on one of the longitudinal ing rollers and a 220 mm span for the support rollers, surfaces with elastic white and black spray paints. with 10 mm diameter ﬁxed loading pins, and are loosely The preparation process required for the produc- connected to the testing machine to allow some hinging tion of the kiln-cast specimens is described in detail in (Fig. 2a). Table 1. Apart from the kiln-cast specimens, the follow- 2nd series of experiments (53 kiln-cast, 5 reference ing industrially manufactured specimens are prepared specimens) and used as a reference: The 2nd set of experiments involves the repeated testing of each glass category and provides accurate displace- • Beams cut out from standard Poesia cast glass ment data. The number of tested specimens per glass bricks, ground and polished to a 30 × 30 × 240 mm category is set to three, which is limited for testing a size brittle material whose strength is by default statistical • Beams from 8/10 mm thick ﬂoat glass plies, adhe- due to the randomness of the occurring ﬂaws in the glass sively bonded with Delo Photobond 4468, ground (Quinn et al. 2009). This study, however, aims to cover a and polished to a 30 × 30 × 240 mm size broad variety of glass types and compare them accord- ing to their ﬂexural behavior, in order to explore which The 600 grit ﬁnishing is set according to ASTM C1161-13. In recycled glass products have further potential for struc- addition, Quinn et al. (2005) observe in their study on Machining tural use. For these tests, a Schenck 100KN displace- Cracks in Ground Ceramics that sintered reaction bonded silicon nitride ﬂexural specimens with 600 grit grinding fail due to mate- ment controlled hydraulic universal testing machine is rial ﬂaws rather than machining damage. This observation can employed, and the specimens are tested in a labora- be extended to glass specimens. tory air environment using a 0.3 mm/min displacement Poesia is the producer of the cast glass bricks for the Crystal rate, which approximately corresponds to a 0.5 MPa/s Houses façade (Oikonomopoulou et al. 2018a). rate. The four-point bending ﬁxtures have a span of This UV-curing acrylate is chosen because it forms a strong bond with the glass surfaces that leads to the monolithic behaviour of the glued sample (Oikonomopoulou et al. 2018a). A slightly faster displacement rate was chosen for the 2nd Under four-point bending, the bonded glass sample is expected to series, with the aim to reduce the total number of DIC images show cohesive failure in the substrate (glass ply) and not delam- per experiment and thus conﬁne the size of the ﬁles produced by ination. the image processing software GOM Correlate to a maximum of 123 450 T. Bristogianni et al. Fig. 2 a Fixture and set-up of 1st series of four-point experi- for the DIC measurement. The metallic strips placed next to the ments. b Fixture and set-up of 2nd series of four-point experi- support pins are cushioning the specimen upon fracture and pro- ments. An LVDT sensor is placed at the middle of the span. The tect the LVDT sensor from damage. No contact occurs between front surface of the specimen is covered with a speckle pattern the specimen and the strips during the bending test 100 mm for the loading rollers and 200 mm for the sup- therefore given the software accuracy of 0.05 pixel, any port rollers, with 20 mm diameter ﬁxed loading pins displacement above 1.57 μm can be captured. (Fig. 2b). To allow for minor adjustments and rota- Flexural strength and Young’s modulus calculation tional movements, the support ﬁxture is placed on a The ﬂexural strength (σ) is computed from the equation semi-circular pin, while the loading ﬁxture is loosely below: connected to the testing machine. In addition, a 1 mm 3 · F · (L − L ) σ = (1) thick silicone rubber strip is placed between each load- 2 · b · d ing pin and the specimen. where F the maximum load, L the support span, L the To measure the displacement of the beam due to load span, b the beam’s width and d the beam’s height. bending, two methods are employed: 1) a Linear Vari- The calculation of Young’s (E) modulus is per- able Differential Transformer (LVDT) displacement formed by correlating the force data obtained from the sensor (Solartron AX 2.5 Spring Push Probe calibrated Schenck machine with (1) the maximum displacement toa0.5 μm accuracy) is placed under the middle point from the LVDT sensor and (2) the maximum displace- of the lower surface of the beam (measuring the point ment from the DIC analysis (Fig. 3). of maximum displacement), and 2) a 2D-DIC measure- ment, using a high-resolution (50.6MP) Canon EOS It should be noted that due to the ﬁxed loading pins, a sys- 5Ds camera that takes one picture per second of the tematic positive error may occur due to a frictional constraint of speckled surface of the beam. The pictures of the 2D- μ · F/2 occurring at each pin, with μ being the coefﬁcient of friction (Quinn et al. 2009). This force creates a counteracting DIC measurement are analysed using the GOM Corre- moment of μ·F·d/2, thus the above equation should be rewritten late software. One image pixel corresponds to 31.5 μm, as: 3 · F · (L − L − μ · d) σ = (2) Footnote 8 continued 2 · b · d 25 Gigabytes. Both the 1st and 2nd series displacement rates are below the rate of stress increase of 1.1 ± 0.2MPa/s indicated by Assuming a moderate μ = 0, 3, the systematic error could be ASTM C158-02. A displacement controlled rate is favoured over of magnitude 8.2% for the 1st series of experiments and 9% force controlled, to avoid the crashing of the specimen upon fail- for the 2nd. However, due to insufﬁcient data regarding the μ ure, but also to allow for potential pop-ins (slight crack arrests) at value, the ﬂexural strength is not corrected in this study, and maximum force, when the crack front interacts with an interface the reader should take into account the possibility of an error of encountered in the glass mesostructure. approximately the aforementioned magnitude. 123 Investigating the ﬂexural strength of recycled cast glass 451 Fig. 3 Analysis of the displacement in y axis, using GOM Correlate software. The maximum displacement due to bending at point A is calculated by subtracting the total displacement at point A from the average displacement at point B and Given that the cross section of the beam in relation 3 Results to the ﬁxture spans results in a relatively stiff struc- tural element, a shear deﬂection should be accounted 3.1 Defect evaluation for kiln-cast specimens to the total vertical deﬂection. The bending and shear deﬂection at mid-span with respect to the beam point The ﬂaws occurring in the surface and bulk of the pro- above the support pins, and for a 1:2 four-point bending duced glass specimens are qualitatively documented ﬁxture ratio, are deﬁned by the formulas below: according to type and cause. Aim is to correlate the defects found to the glass source used and followed casting and post-processing procedure, and to subse- L−L quently assess their contribution to the specimens’ ﬂex- 11 · ΔF · Δl = (3) Bending_mid ural strength. The casting related defects are catego- 12 · E · b · d rized in: L−L ΔF · Δl = (4) shear_mid 2 · G · b · d 1. Crystalline Inclusions 2. Glassy inhomogeneities (cord/ream) where 3. Gaseous inhomogeneities (bubbles) G = (5) An overview of the defect categories and their causes is 2 · (1 + v) found in Fig. 4, based on which a documentation of the Adding the two segments of vertical deﬂection and observed ﬂaws per glass type is presented in Table 2. solving towards the Young’s modulus, it is con- cluded: ⎛ ⎞ L−L i 11 11 · The quantitative analysis of the level of inhomogeneities in ΔF 2 (L − L ) · (1 + v) ⎜ i ⎟ E = · ⎝ + ⎠ cast glass specimens of considerable cross section- and thus Δl 12 · b · d 2 · b · d total_mid multiple layers of defects versus a thin-walled glass—is a com- plex process that involves several different testing methods (e.g. (6) Computed Tomography Scanning to detect and measure density differentials, 3-dimensional Imaging Real-Time Polarimetry to deﬁne the location and shape of cord, etc). This analysis is kept 10 out of the scope of this study as the main aim is to ﬁrstly identify For the Young’s modulus calculation, the Poisson ratio of the type and location of critical ﬂaws that require future attention, v = 0.22 of soda lime silica glass is used. Although among the and thus quantitative documentation. tested glasses there may be a ± 0.02 deviation to this value, this has a negligible effect on the results. Categorization based on Bartuška (2008). 123 452 T. Bristogianni et al. Fig. 4 Categorization and causes of the defects encountered in the kiln-cast glass specimens Table 2 Evaluation of kiln-cast specimens 123 Investigating the ﬂexural strength of recycled cast glass 453 Table 2 continued 123 454 T. Bristogianni et al. In more detail, the cause of these defects is associ- “ Borosilicate mix Maltha” specimens. ated with one or more of the following manufacturing More speciﬁcally, the “Float combo” spec- 13,14 stages : imens were cast by employing a compi- lation of ﬂat glass shards (of approx. 20– 50 mm width) provided by Maltha Recy- I. Raw Material. cling. This ﬂat glass compilation is rejected A. Contamination. from the recycling stream as the erro- i. Coatings. neous deposition of glass ceramic plates Several “ﬂat” defects are observed in kiln- (e.g. cooktops) in the ﬂat glass collection cast specimens from ﬂoat glass cullet container—an often encountered covered with enamel paint or ceramic phenomenon—renders the entire container frit, due to the insufﬁcient melting of unsuitable for recycling. The XRF and the coatings (Fig. 5). The XRF analyses XRD analyses of characteristic pieces from of two characteristic coatings (Table 3) the ﬂat glass compilation sample (Table 4; show compositions rich in high melting- Figs. 8b, 9) place the contaminants in the point metal oxides and in particular in commercially applicable lithium alumi- chromium(III) oxide (melting point of nosilicate glass ceramics system, which is Cr O is 2435 C, NIH Database). The X- 2 3 characterized by the close to zero thermal ray diffraction (XRD) analysis of kiln- expansion coefﬁcient (Höland and Beall cast glass samples (Fig. 6) shows in these 2020). The very low thermal expansion cases the presence of eskolaite (mineral coefﬁcient (CTE) contrasts with the typ- name of chromic oxide). −6 ◦ ical 9.5 × 10 /K (at 20–300 C) of ﬂoat ii. Minor compositional variations. glass (Shelby 2005), leading to unavoid- Minor compositional variations lead to able cracking. However, the reduction of glassy inhomogeneities such as cord and the ﬂat-glass compilation sample’s parti- colour streaks. Some examples with heavy cle size (ﬁne cullet or powder), could min- striation are identiﬁed in the “Float combo” imize the strains in the ﬁnal cast product, and “Lead CRT” (Fig. 7) samples. and therefore this strategy requires further iii. External contaminants. investigation. In this category, the presence of glass ceramics or chemically different families Traces of metal, clay or stone lead to crys- of glass in the cullet (not detectable by talline inclusions of a maximum of 2 mm eye, e.g. aluminosilicate shards in borosili- size, but these are tolerated by the glass cate or soda lime silica cullet)), is the most network (Figs. 10, 11). However further critical, leading to specimens which frac- study is required to identify the crystalline ture upon cooling, due to strains caused by inclusions (employing scanning electron thermal expansion variations. This is expe- microscopy) and to test if their role remains rienced in the “Float combo” (Fig. 8a) and neutral when the glass is subjected to tem- perature gradients. B. Cullet size and shape. 13 In the addressed glass viscosity range, the In Table 2, ﬂaws caused during stages III.B. and IV (post- geometry of the cullet is often reﬂected in stri- processing and handling ﬂaws) are not mentioned as they are not linked to the material and its casting method, but are rather ations and/or three-dimensional bubble veils arbitrary and only relevant to the fracture analysis of each speciﬁc in the ﬁnal glass component. In cases of very specimen. ﬁne cullet (e.g. “Car Windshields” samples) The following microscope images were made using a Keyence this geometry is not distinguishable, and a VHX-5000 or VHX-7000 Digital Microscope. rather high content miniscule bubbles prevails All XRD analyses in this work were conducted using a (Fig. 12). Bruker D8 Advance diffractometer, Bragg-Brentano geometry and Lynxeye position sensitive detector. 123 Investigating the ﬂexural strength of recycled cast glass 455 ◦ ◦ Fig. 5 a Microscope image of a “Oven doors, 1120 C” kiln-cast image of a “Car windshields, 1120 C” kiln-cast glass with crys- glass with ﬂat crystalline inclusions, cord, colour streak (due to talline inclusions and bubbles partially molten coating material) and bubbles. b Microscope Table 3 Coating composition Coating type Glass source Composition (wt%) SiO Bi O C O CuO PbO NaOTiO Fe O Al O CdO ZnO 2 2 3 2 3 2 2 2 3 2 3 Black enamel Enamel ﬂoat, AGC 30.7 28.6 19.4 10.2 3.3 4.2 1.3 Black frit Oven door, Coolrec 33.2 22.5 10.6 12.7 7.2 2.2 3.8 3.6 2.9 XRF measurements conducted with a Panalytical Axios Max WD-XRF spectrometer by Ruud Hendrikx. The absolute wt% obtained by the XRF measurements may not be entirely accurate in the case of thin coatings, due to the extremely small thickness of the coating Fig. 6 XRD pattern of kiln-cast “AGC Float with black enamel” glass at 1120 C. Thesampleisata large extent amorphous (black curve) yet it presents some sharp crystalline peaks (coloured sticks) 123 456 T. Bristogianni et al. Fig. 7 “Lead CRT, 870 C” specimen containing intense cord (seen as wavy lines) and bubbles ◦ ◦ Fig. 8 a Fractured “Float combo, 1120 C” specimen due to 1120 C for 10 h. This behavior suggests a lithium aluminosil- glass ceramics contamination. b Glass ceramic shards encoun- icate β-quartz solid solution phase in the transparent condition tered in the ﬂat glass compilation sample. The left column shows that transforms to β-spodumene during heat-treatment at temper- the shards in the “as-received” transparent condition, whereas the atures above 1000 C. The larger crystals in the later condition right column shows their opaque version after heat-treatment at scatter the light and lead to opacity (Shelby 2005) Fig. 9 XRD pattern of the yellow transparent glass ceramic depicted in Fig. 8b, in the as received condition (a) and after heat-treatment at 1120 C for 10 h (b). The crystal structure in b is similar to β-spodumene, yet the material presents multiple phases 123 Investigating the ﬂexural strength of recycled cast glass 457 Table 4 Chemical composition, crystal phase and CTE of typical lithium aluminosilicate glass ceramics, compared to the tested glass ceramic samples, the cast “Float Combo Maltha” specimen and a typical window glass *Lithium is a light element that cannot be detected by the XRF analysis and therefore the percentage corresponding to lithium oxide is reﬂected to a higher content of silica dioxide. According to the bibliography, the presented composition should have a 2-3% lithium oxide content and a lower silica dioxide content by 2-3% **A lower than 3% lithium oxide content is expected in the chemical composition  XRD measurements conducted by Ruud Hendrikx (TU Delft, 3me) using a Bruker D8 Advance diffractometer, Bragg–Brentano geometry and Lynxeye position sensitive detector  XRF measurements conducted with a Panalytical Axios Max WD-XRF spectrometer by Ruud Hendrikx (TU Delft, 3me);  Montazerian et al. (2015);  Songhan Plastic Technology Co., Ltd.;  Schott (2015b);  Höland and Beall (2020);  Shelby (2005);  Chyung (1977);  Brennan (1979);  Campbell and Hagy (1975) Fig. 10 a Microscope image of “Lead CRT, 870 C” specimen, containing undissolved blue particles of—most probably—cobalt oxide. b The variable inclusions in the “Borosilicate mix Coolrec, 1120 C” specimen are tolerated by the glass network 123 458 T. Bristogianni et al. Fig. 11 Crystalline inclusions and bubbles detected in the bulk of a “Schott DURAN tubes 1120 C” specimen, viewed through cross-polarized light. Although some inclusions e.g. the depicted 62.5 μm stone, induce stress to the surrounding glass, this is well tolerated within the 30 × 30 mm glass cross section Fig. 12 The size, shape, and arrangement of the cullet, in com- (image width ≈ 30 mm), b “Wertheim, 820 C” (image height ≈ bination with the forming temperature, lead to organized (a), ran- 30 mm), c “Car windshields, 1120 C” (image height ≈ 30 mm) dom visible (b) and random non-traceable (c) meso-structures in the glass component. a “AGC ﬂoat with black enamel, 1120 C” II. Glass forming. B. Forming temperature and corresponding vis- cosity in relation to dwell time. The top temperature affects the level of homog- A. Cullet arrangement in the mould. enization and the content of air-bubbles. All This is relevant with the geometry of the cul- samples present miniscule bubbles due to the let (I. B) in combination with the ﬁring sched- relatively low forming temperatures. In addi- ule and corresponding viscosities of the formed tion, the “cage” principle describing the mixing glass (II. B, C). A deﬁned cullet shape and high of dense liquids is applicable in this case, mean- viscosity can lead to organized meso-structures ing that most of the molecules corresponding to composed of bubble veils (Figs. 13b, 14), cord an initial cullet piece will remain in the same or crystallized interfaces which result in a more position in relation to their neighboring cluster predictable failure pattern (Fig. 13). Such orga- of molecules (cullet piece). The level of diffu- nized structures also help in distinguishing the 3.5 sion is increased when a viscosity of 10 dPa s role of these defects when present at the glass magnitude is reached, but it does not in any case surface or in the bulk. 123 Investigating the ﬂexural strength of recycled cast glass 459 Fig. 13 Kiln-cast experiments with Schott DURAN borosilicate rods of 24 mm diameter forming 50 mm cubic samples. a Crystallized ◦ ◦ hexagon structure, engineered at 970 C. b Bubble-veil hexagon structure engineered at 1120 C Fig. 14 Bubble veil observed in a “Schott DURAN tubes 1120 C” specimen. The maximum bubble diameter is less than 1 mm, while the majority of the bubbles has a diameter below 0.2 mm lead to a fully mixed glass in the given dwell cristobalite and devitrite (Figs. 16, 17). Crys- time (see Figs. 12, 13). tallization is favoured because the samples C. Firing schedule in combination with tempera- are formed below their liquidus point (T is ture differentials in the kiln that promote crys- around 1080 C for the speciﬁc ﬂoat glass, and ◦ 16 tallization. around 1200 C for the speciﬁc borosilicate ) This is particularly applicable for the ﬂoat and yet reaching a low enough viscosity that kineti- borosilicate glass samples formed at 970 C. In cally allows nucleation. Nucleation starts at the these samples, the complete interface between interfaces, as there, a local compositional vari- each cullet piece is crystallized. According to ation occurs due to the volatilization of alkali the XRD analysis (Fig. 15), the borosilicate The liquidus point of glasses T is found around a viscosity samples develop b-cristobalite crystals, while of 10 dPa s, and is estimated from the chemical composition of the ﬂoat glass samples wollastonite 2M, b- the given glasses according to Fluegel (2007a). 123 460 T. Bristogianni et al. Fig. 15 XRD patterns of ﬂoat glass (left) and Schott DURAN borosilicate rods (right) fused at 970 C Fig. 16 Microscope images of the crystallized interface of the “Float 10 mm fused 970 C” samples (fractured surface). The parallel needle-like form of the crystals refers to devitrite Fig. 17 Crystallized interface of the “Schott DURAN borosilicate rods, fused at 970 C” samples. a Microscope image showing a split interface due to fracture. b Water permeability of the crystallized interface (image height ≈ 30 mm) 123 Investigating the ﬂexural strength of recycled cast glass 461 and boron (in the case of the borosilicate glass). E. Quenching rate to the annealing point. However, depletion of such elements may lead In this study a lower quenching rate of to unstable local compositions, as observed in − 160 C/h is adopted in comparison to the the crystallized layer of the borosilicate sam- abrupt quenching followed in industrial glass ples, which proves porous and water-absorbing casting. The experimental results show that (Fig. 17). Apart from the “engineered” crystal- this rate is sufﬁcient to prevent crystallization. lized structures described above, the tempera- However, attention is raised to the fact that a ture conditions and ﬂuctuations within the kiln slower cooling rate may intensify the level of can also provoke local and random crystalliza- polymerization of the glass network and lead tion in the form of stones, at locations of compo- to a denser glass (Ito and Taniguchi 2004). sitional alteration. Local variations in the com- Although this is not experimentally proven in position can be caused by contaminants in the this study, it remains a possibility to be taken raw material, contact with the mould material, into account. volatilization of compounds, and gas bubbles. F. Annealing scheme. Therefore, such stones are not only found in A conservative annealing scheme has been specimens produced from evidently contami- used, thus the residual stresses detected in the nated cullet (e.g. “Car windshields” samples), samples using cross-polarized ﬁlters are negli- but also in more pure specimens (e.g. “Fully gible and do not seem to compromise the ﬂex- tempered (FT) ﬂoat” samples). ural strength. Regarding the samples cut out D. Reaction with mould surface. from the standard Poesia glass bricks, these During the kiln-casting process (at the stud- do have minor residual stresses, which is also ied viscosity range), the glass in contact with seen by the fringe order in the isochromatic pat- the silica/plaster investment mould, forms a tern obtained by an Ilis StrainScope Flex cir- thin crystallized interface, that can be eas- cular polariscope (Fig. 20), and also suggested ily removed by the described post-processing strongly by the tendency of this glass to chip methodology (Fig. 18). However, of particular during post-processing. interest are defects caused by the interaction of III. Post processing. the mould with the glass that are deep enough to remain upon grinding (Fig. 19). These can A. Inadequate removal of existing ﬂaws. be, for example, stones of approx. ø 1–2 mm As also discussed at point II.D, not all sur- created from loose mould material that acciden- face ﬂaws can be completely removed by tally got incorporated in the glass melt. Another post-processing (Fig. 21a). In this category characteristic ﬂaw occurs due to the friction of defects, the exposure during grinding of of the mould surface that obstructs the com- bubbles trapped in the glass bulk should be plete fusion between the cullet pieces. As a included. This results in stress concentrating result, localized or networks of infolds appear semi-circular intrusions of sharp edges at the at the glass surfaces, which can also encapsulate glass surface that reduce the strength. In addi- mould material. Upon grinding, the tip of these tion, since bubbles can offer favourable condi- ﬂaws may remain at the glass surface, and is tions for the formation of crystals in their inte- observed in depths up to 5 mm. Lastly, only one rior, the exposure of such gas-pockets at the sur- case is observed where the glass bonds to the face bare the additional risk of stone exposure mould surface and breaks during cooling due to (see Figs. 19b, 21b). thermal expansion variations (sample “Borosil- B. Introduction of new ﬂaws. icate mix Maltha”). The introduction of new scratches from “rene- 17 18 The designation “Fully tempered” refers to the cullet used for In this study, quenching may last even 4 h and takes place these samples, which originates from shattered fully tempered within the kiln, which is inherently different from the quenching ﬂoat glass panels. The ﬁnal kiln-cast components are annealed at atmospheric conditions during hot-pouring of glass that lasts and thus not tempered. only several minutes. 123 462 T. Bristogianni et al. Fig. 18 Surface reaction to the mould material. a Side surface “Oven doors, 1120 C” specimen, as released from the mould. of a “Car Windshields, 1120 C” sample, as released from the The white zones are crystalline formations from the reaction of mould. Improper fusion of the cullet, inclusions from the mould the glass coating to the mould material. Improper fusion of the material and stone formations are observed. b Side surface of an cullet is also observed, as well as mould material inclusions Fig. 19 a Infold with stone inclusions from the reaction of the 1120 C” specimen. Note that the crystalline inclusion from the glass to the mould material, in the ground surface of an “Oven reaction to the mould, are not only situated at the bottom surface, doors, 1120 C” specimen (image height ≈ 30 mm). b Micro- but extend to the bulk as well scope image of the fracture origin of a “Schott DURAN tubes gade” abrasive grits (Quinn 2016) is mainly observed in glasses with lower hardness, in this study particularly the “Leerdam Lead” samples. Chipping is mainly occurring in the cut-out standard Poesia samples, as discussed in II.F. All samples present the risk of micro-cracking during coarse grinding that is not sufﬁciently removed in the later stages of grinding and pol- ishing. IV Handling. Fig. 20 Isochromic fringes observed via an Ilis StrainScoep Flex circular polariscope in a standard Poesia cast glass brick. The A series of handling ﬂaws (chippage, cleavage, depth of the depicted sample is 10 cm percussion cone, point contact) randomly occur in some of the specimens. The response of the cast 123 Investigating the ﬂexural strength of recycled cast glass 463 Fig. 21 a Clustering of surface bubbles and stone inclusions ﬂaw. b Microscope image of a fractured “Schott DURAN tubes ◦ ◦ at the bottom surface of an “Oven doors, 1120 C” specimen 1120 C” specimen, showing a bubble in proximity to the surface that were not removed after grinding, form the strength limiting and stone formations originating from the bubble interior Fig. 22 Overview of tested specimens specimens to handling damage versus that of indus- 3.2 Four-point bending tests trially produced glass (e.g. ﬂoat, extruded rods) requires further investigation, yet the more pure cast The results of both series of experiments are presented specimens are not observed to be more susceptible in Figs. 22, 23, 24, 25 and 26 and Tables 5, 6, 7 and 8. than standard glass products. However, attention The data from the ﬁrst series is mainly used for a ﬁrst should be drawn to the more contaminated glass general guidance and as a conﬁrmation of the second samples, as occasional large defects at the surface series, which is the main focus of this study. It should (> 2 mm) amplify the effect of an impact. be stressed that the number of tested specimens per category is limited, and thus the presented results are only indicative and not sufﬁcient for deriving statistical conclusions. 123 464 T. Bristogianni et al. Fig. 23 Flexural strength results of 1st series of four-point bending experiments Fig. 24 Flexural strength results of 2nd series of four-point bending experiments 123 Investigating the ﬂexural strength of recycled cast glass 465 Fig. 25 Force versus Displacement graph. The displacement is measured from the DIC analysis Fig. 26 Comparative graph of the Young’s modulus measured by the LVDT sensor during the 2nd series of four-point bending experiments Although the ﬁrst and second series differ in the ﬁx- the two tests are aligned. More speciﬁcally, the sam- ture set-up (span, roller radius, connection detail to uni- ples of the 1st series that are cast at 1120 C (“FT versal testing machine) and the sensitivity of the testing Float”, “Schott DURAN 24 mm rods”, “Oven doors machine (10KN max. applied load for the machine used Coolrec”) score within the same ﬂexural strength range in 1st series versus 100KN for the 2nd), the results of (40–50 MPa), the fused samples at 970 C are signiﬁ- 123 466 T. Bristogianni et al. Table 5 Results of 1st series of four-point bending experiments concerning the kiln-cast beams 1st four-point bending experiment: Kilncast glass beams 30 × 30 × 240 mm, 110/220 mm supports Glass type Specimen Forming No. of tested Flexural strength Average ﬂexural description temperature ( C) specimens (MPa) strength (MPa) Minimum Maximum Soda lime silica Fully tempered 1120 3 40.9 46.5 43.9 (ﬂoat glass) ﬂoat Fully tempered 970 1 17.9 – ﬂoat Float 10 mm, 3 970 1 9.5 – horizontal layers Float 10 mm, 24 970 1 9.9 – vertical layers Float 10 mm, 3 970 1 9.4 – vertical layers Oven doors 1120 1 46.1 – Borosilicate DURAN 1120 2 44.2 49.5 46.8 rods × 10 vertical DURAN 970 1 15.5 – rods × 10 vertical DURAN rods × 2 970 1 18.5 – vertical Table 6 Results of 1st series of four-point bending experiments concerning the reference beams 1st four-point bending experiment: reference beams (240 mm length), 110/220 mm supports Glass type Specimen Width No. of tested Flexural strength Average ﬂexural description specimens (MPa) strength (MPa) Minimum Maximum Soda lime silica Float 8 mm, 3 30 1 48 – horizontal layers glued with DELO Float 8 mm, single pane 30 1 43.7 – Float 10 mm, single 50 3 43.8 64 54.8 pane cantly weaker (10–20 MPa) while the pure single pane of these specimens, which could not be easily removed ﬂoat samples have a slightly better performance (aver- by post-processing. age ﬂexural strength of 55 MPa). This performance In Fig. 24, depicting the ﬂexural strength of the ranking and value range coincides with the results of the second series of specimens, three main zones can 2nd series apart from the case of the fused ﬂoat samples be observed: specimens of a ﬂexural strength below at 970 C, where a noticeably low ﬂexural strength is 30 MPa, between 30 and 55 MPa—where most sam- reported (< 10 MPa). This is attributed to the one-off ples are located, and between 55 and 75 MPa. In all occurrence of a network of micro-cracks at the surface specimens, crack initiation starts at the bottom sur- 123 Investigating the ﬂexural strength of recycled cast glass 467 Table 7 Results of 2nd series of four-point bending experiments concerning the kiln-cast beams 2nd four-point bending experiment: Kilncast glass beams 30 × 30 × 240 mm, 100/200 mm supports Glass type Specimen Forming temperature No. of tested Flexural strength Average ﬂexural Average E modulus description ( C) specimens (MPa) strength (MPa) (GPa), LVDT calculation Minimum Maximum Soda lime silica Fully tempered ﬂoat 1120 6 33.5 50.8 43.7 59.3 (ﬂoat glass) Float 10 mm, 3 970 2 41.1 49.9 45.5 59.7 horizontal layers Float 10 mm, 24 970 3 27.9 37.7 33.3 58.6 vertical layers Float dark blue 1120 3 60.7 65.7 62.9 62.3 Soda lime silica Float combo 1120 1 46.5 – 61.8 (ﬂoat glass) with contamination Oven doors 1120 3 29.9 43.8 37.5 58.3 Car windshields 1120 3 35.2 45.9 41.1 59.8 Float with black 1120 3 36.4 50.1 41.7 60.7 enamel, 60 layers Modiﬁed soda lime Poesia standard cast 1070 3 61.3 72.1 66.5 61.1 brick 468 T. Bristogianni et al. Table 7 continued 2nd four-point bending experiment: Kilncast glass beams 30 × 30 × 240 mm, 100/200 mm supports Glass type Specimen descrip- Forming temperature No. of tested Flexural strength Average ﬂexural Average E modulus tion ( C) specimens (MPa) strength (MPa) (GPa), LVDT calculation Minimum Maximum Borosilicate DURAN tubes 1120 5 30 50 42.5 52.4 DURAN rods × 10 1120 3 34.3 50.6 44.1 53 vertical DURAN rods × 10 1070 3 24.5 36.7 30 50.9 vertical DURAN rods × 10 970 3 10.1 15.9 12.4 36.9 vertical DURAN 24 mm 970 1 19.4 – 41 rods, honeycomb Borosilicate mix 1120 1 66.9 – 59.2 Coolrec Wertheim (C-glass) Wertheim pellets 820 3 41.3 61.1 52.2 62.6 Wertheim pellets 900 1 63.4 – 64.6 Barium–strontium Barium CRT front 870 3 42.4 56.1 51.2 58 silicate panel Potash–lead silicate Lead CRT funnel 870 1 33.3 – 56.1 Lead glass 820 2 32.5 38.1 35.3 49.8 The LVDT calculation results in a lower than expected Young’s modulus by approximately 15%, due to sensor errors. The provided Young’s modulus data are only for comparison between the different glass types Investigating the ﬂexural strength of recycled cast glass 469 Table 8 Results of 2nd series of four-point bending experiments concerning the reference beams 2nd four-point bending experiment: reference beams 30 × 30 × 240 mm, 100/200 mm supports Glass type Specimen No. of tested Flexural strength Average ﬂexural Average E description specimens (MPa) strength (MPa) modulus (GPa), LVDT calculation Minimum Maximum Soda lime silica Float 10 mm, 3 2 44 52.5 48.3 49.7 horizontal layers glued with Delo 4468 Soda lime potash Poesia standard 3 42.9 59.3 50.4 59 borosilicate cast glass brick, cut & polished The LVDT calculation results in a lower than expected Young’s modulus by approximately 15%, due to sensor errors. The provided Young’s modulus data are only for comparison between the different glass types Fig. 27 Side view of kiln-cast specimens fractured during the 2nd series of four-point bending tests. Note that the primary crack starts perpendicular to the beam’s long axis and then splits in the case of medium/large accumulated elastic energy (prior to cracking), or propagates as one crack in the case of low energy (e.g. crystallized specimens). At the top (compressive zone), the crack forms compression curls face (or at very close proximity), at the area between gories: stones, crystalline interfaces, surface bubbles, the support pins (zone of maximum tensile stress, and machining damage. The size of the fracture mirror see Figs. 27, 28). As a general trend, glass speci- is measured in a selection of specimens (Fig. 30) and mens produced at lower viscosities and from purer plotted against the ﬂexural strength σ (Fig. 31) based cullet are found at the top zone of the ﬂexural on “Orr’s equation” (Quinn 2016): strength graph, while specimens with obvious strength- limiting ﬂaws exposed at the bottom surface fail at low σ = √ (7) values. An overview of the main fracture origins is presented where R corresponds to the mirror radius (in this study in Fig. 29, summarizing the most critical defect cate- the mirror size extending to the mist-hackle boundary at 123 470 T. Bristogianni et al. Fig. 28 Graph depicting the location of the fracture origin of the 2nd series specimens at the bottom surface. Note that fracture origins found at the two long edges are usually relatedtomachiningﬂaws Fig. 29 Mirror surfaces of fractured specimens (2nd series of experiments) depicting the main defect categories responsible for catastrophic failure. The reported ﬂexural strength is linked to the type of defect but also to its size the bottom surface of maximum tension is measured) of the width of the critical ﬂaw is, as expected, respon- and A is the characteristic mirror constant per glass sible for the decrease of the ﬂexural strength (Fig. 32). composition. The higher strength specimens seem to fail mainly Typically, the larger the failure stress is, the smaller from machining ﬂaws, whereas stones or crystalline the encountered fracture mirror will be. The increase interfaces are responsible for the fracture of the lower 123 Investigating the ﬂexural strength of recycled cast glass 471 Fig. 30 Measuring example of the fracture mirror and defect size stress gradient along the height of the sample (due to loading in at origin. All measurements are conducted employing a Keyence bending), the mirrors appear elongated in this direction, or may VHX-7000 Digital Microscope and with the fractured surface even be incomplete. Therefore a measurement along the bottom positioned perpendicularly to the microscope’s optical path. To surface is preferred. Moreover, extended ﬂaws at the surface or obtain the mirror radius, the diameter of the mist-hackle bound- machining damage can cause the one-sided elongation of the ary at the bottom surface line (maximum tensile stress) is mea- mirror, and thus the measurement of the diameter instead of the sured and then divided by half. This method is chosen as not radius is opted all mirrors are found semi-circular. More speciﬁcally, due to the Fig. 31 Flexural strength versus 1/ R graph for a selection of tionship of different glass compositions is reported, more than glass specimens. In general, the higher the strength, the smaller one mirror constants A are applicable, and thus the data are not the mirror size. However, since the mirror size to strength rela- all corresponding in one line (e.g. “Wertheim pellets” samples) 123 472 T. Bristogianni et al. Fig. 32 Flexural strength versus critical ﬂaw width (at fracture stones tend to fail at lower strength values than the purer sam- origin) for a selection of glass specimens. The strength reduces ples that fail from post-processing and handling ﬂaws with the increase of the ﬂaw size, and specimens with surface strength specimens. Yet, the type, size, quantity and 4 Discussion location of ﬂaws alone cannot justify why some glass samples score lower than others. The structural perfor- The ﬂexural strength of the cast glass specimens is mance of each glass type needs to be reviewed as con- conjointly related to their chemical composition and jointly dependent on the chemical composition of the inherent defects. To comprehend in which cases the glass as well as its inherent defects (see Sect. 4). Also, ﬂaws are the strength limiting factor and when the fracture load uncertainty may be applicable due to envi- mechanical properties related to the composition have ronmentally assisted slow crack growth (Quinn et al. a determining role, the interpretation of the results 2009), as the applied loading rate is slower (approx. by is structured in the following categories: a. Non- half) than the suggested rate by the ASTM C158-02 contaminated glass specimens, b. Contaminated vs. guideline. The effect of slow crack growth should be non-contaminated glass, c. Non-contaminated homo- further experimentally investigated in a broader range geneous glass specimens vs. with crystallized inter- of testing speeds. faces, and d. Reference specimens. In this manner, the Regarding the Young’s modulus, the calculation defects are categorized and isolated so their effect can conducted based on the LVDT data results in values be studied with more clarity, while the absence of over- that are approximately 15% lower than those found in ruling ﬂaws (in the case of the pure samples) highlights literature. This is considered a systematic error and is the effect of the chemical composition. attributed to the quality of the sensor. However, in each (a) Non-contaminated glass specimens triplet of tested glass type, there are matching E values reported. In addition, the stiffness relationship between The purest, most homogeneous samples of each glass the different glass families (Fig. 26) is found in accor- family included in this work are selected for compar- dance with the literature (Corning 1979; Campbell and ison (Figs. 33, 34). As these examples contain less Hagy 1975), and speciﬁcally: imperfections, the effect of their chemical composi- tion on their ﬂexural strength is highlighted. Table 9 lists relevant calculated and/or measured physical and E ≤ E Potash Soda Lead-silicate Borosilicate mechanical properties of these glasses, along with data < E < E BaO/SrO-Silicate Soda Lime Silica found in literature for similar compositions. 123 Investigating the ﬂexural strength of recycled cast glass 473 Fig. 33 Average ﬂexural strength to Young’s modulus graph concerning the non-contaminated kiln-cast specimens of the 2nd series ◦ ◦ Fig. 34 Side surface of fractured specimens. a First three spec- 870 C” glass. b From top to bottom: “Wertheim 900 C”, “Poe- ◦ ◦ ◦ imen correspond to “AGC dark blue ﬂoat 1120 C” glass, then sia 1070 C” and three “FT Float 1120 C” specimens the next two are “Leerdam 820 C”, followed by a “Barium CRT Therefore, although the LVDT calculation does not tation Technique, is advised in future testing, to verify provide exact values, it can be reliably used for a com- the reliability of the results. parative analysis between the different glass types. The As seen in Fig. 33, there is an increase in the ﬂex- DIC measurement is utilized to provide more accurate ural strength with increasing Young’s modulus, in data regarding the maximum deformation, and for per- the lead silicate, borosilicate, barium silicate and AGC forming more precise calculations of the E moduli for The graph in Fig. 33 is based on the Young’s modulus cal- a selection of glass samples (see Sect. 4a). Nonethe- culated from the DIC measurements. The reported E modulus less, the coupling of the DIC measurement during 4- is approximately 5% higher than in literature, which could be point bending with a non-destructive testing method for partly related to testing errors and partly to the material itself determining the E modulus, such as the Impulse Exci- and its casting procedure. 123 474 T. Bristogianni et al. Table 9 Measured and calculated properties of the selected (pure) glasses (in bold), and reference glasses of similar composition Glass type Name Composition (wt%) SiO2 B2O3 Na2OK2OCaO MgO Al2O3 PbO Fe2O3 Sb2O3 ZnO BaO SrO Source Soda lime Standard ﬂoat 70–74 12–16 0–0.5 8–13 0–5 0–2 0.01–1.5  silica PPG Starphire 74.6 13.3 8.9 3 0.04  (low iron) FT Float 75.4 12.4 7.6 4 0.4 0.09  AGC blue 73.1 12.8 8.1 4 0.9 0.76  Modiﬁed Poesia glass 72.1 2.5 15.9 1.9 6.1 0.06 0.3 0.9 [5, 8] Soda Lime Borosilicate Corning 7740 80.6 13 4 2.3  Pyrex Schott 81 13 4 2  DURAN C-Glass Johns Manville 63.5 5.5 14.6 1 6 3 5.5 0.1  753 C-glass ﬁbers Wertheim 63.8 5.5 11.8 3.2 6.4 3.7 5.2 0.06 0.08 [5, 14] glass SrO/BaO Corning 9068 63.2 7.1 8.8 1.8 0.9 2 2.3 0.4 2.4 10  Silicate Philips CRT 61.6 7.2 6.8 1.1 0.3 2.3 0.1 8 8  panel Potash-lead- Corning 0120 55 3 4 9 2 27  silicate Leerdam glass 57.7 3 9 28.7 0.8 0.6  Glass type Name Annealing Density Poisson’s Knoop Molar volume APF calculated G Total E(GPa), E(GPa) E (GPa) from E (GPa) from Average 3 3 point (g/cm ) ratio Hardness V (cm /mol), based on Dissociation calculatedˆ from LVDT DIC Flexural 10 dPa s KHN100 calculated Pauling’s ionic energy literature measurementˆˆ measurementˆˆˆ Strength (2nd ◦ 3 ( C) radii (kJ/cm ), 4-PB test) calculated (MPa) Soda lime Standard ﬂoat 525–545  2.480–2.520  0.22–0.23  550  70–75  silica PPG Starphire 547  2.510  0.22  448  23.55 0.5492 64.03 70.33 73.1  (low iron) FT Float 553  2.466  23.92 0.5413 64.84 70.19 59.3 72.7 43.71 AGC blue 550  2.492  23.81 0.5436 64.87 70.52 62.3 76.5 62.9 c c Modiﬁed Poesia glass ≈ 520  2.486  24.65 0.5471 61.84 (IV) 67.67 (IV) 61.1 75.8 66.5 Soda Lime Borosilicate Corning 7740 560  2.230  0.20  418  64  Pyrex Schott 560  2.230  0.20  480  27.53 0.5383 64.13 (B O 66% 69.04 (B O 66% 63  52.4 66.8 42.45 2 3 2 3 DURAN III, 34% IV) III, 34% IV) Investigating the ﬂexural strength of recycled cast glass 475 Table 9 continued Glass type Name Annealing Density Poisson’s Knoop Molar volume APF calculated G Total E (GPa), E(GPa) E (GPa) from E (GPa) from Average 3 3 point (g/cm ) ratio Hardness V (cm /mol), based on Dissociation calculatedˆ from LVDT DIC Flexural 10 dPa s KHN100 calculated Pauling’s ionic energy literature measurementˆˆ measurementˆˆˆ Strength (2nd ◦ 3 ( C) radii (kJ/cm ), 4-PB test) calculated (MPa) a c c a C-Glass Johns Manville 527  2.520  0.27  0.5586 66.01 (IV) 73.74 (IV) 68.9  753 C-glass ﬁbers c c Wertheim 550  2.502  24.56 0.5555 66.38 (IV) 73.75 (IV) 63.6 79 54.98 glass SrO/BaO Corning 9068 503  2.696  0.5456 61.00 66.56 69.6  Silicate Philips CRT 530  2.766  25.24 0.55 62.22 68.20 58 73.5 51.19 panel c c Potash- Corning 0120 435  3.050  0.22  382  0.5484 60.17 (IV) 65.99 (IV) 60  Lead- Silicate Leerdam glass 465  3.031  26.5 0.5288 58.35 61.71 49.8 64.9 35.29 Sources  Quinn and Swab (2017)  Shelby (2005)  Martienssen and Warlimont (2005)  Specialty Glass Products  XRF measurements conducted by Ruud Hendrikx  Calculated using viscosity model by Fluegel (2007a)  Calculated using density model by Fluegel (2007b)  Personal correspondance with Poesia  Friedrich & Dimmock Inc.  Corning (1979)  Schott (2015a, b)  Schott (2017)  Abrisa Technologies (2014)  Campbell and Hagy (1975)  Matweb  ASM International (1995)  Thompson (1980)  Gregory (1998) Properties corresponding to a generic c-glass ﬁber Calculated based on data from Inaba et al. (1999) (III) Corresponds to B O with coordination number = 3, (IV) to B O with coordination number = 4. The coordination number ratio for the Schott DURAN glass is calculated 2 3 2 3 using the formulas proposed by Yun and Bray (1978). For the Poesia, Wertheim and Corning 0120 glass, a 100% fourfold coordination is assumed given the high alkali/boron ratio (Priven 2000; Zhdanov and Shmidel’ 1975). ˆ Calculated using V values from Makishima & Mackenzie (1972), and G values from Inaba et al. (1999) i i ˆˆ The LVDT calculation results in a lower than expected Young’s modulus by approximately 15%, due to sensor errors. The provided data are only for comparison between the different glass types. ˆˆˆ The reported E modulus is approx. 5% higher than in literature, partly due to testing errors and partly due to the material itself and its casting procedure 476 T. Bristogianni et al. dark blue ﬂoat glass samples. The increase in strength silicate (SLS) glasses will have the highest strength. is attributed to the increase of the average bond strength Also in accordance with the literature, the BaO con- and atomic packing density of the glass network. This taining silicate glass has a lower ﬂexural strength and is related to the Young’s modulus by the equation below Young’s modulus than the CaO silicates but higher than (Makishima and Mackenzie 1973): the lead silicates (Volf 1984; Corning 1979). However, the Young’s modulus alone cannot justify the devia- tion from the linear E/strength relationship that present E = 2 · C · G (8) g t the Poesia, Wertheim, and FT ﬂoat glass samples, and further explanation is required per glass type. where C is the atomic packing density (also mentioned The Poesia glass is a modiﬁed soda lime glass with as Atomic Packing Factor, APF), and G the total disso- a decreased forming temperature compared to conven- ciation energy per unit volume. Based on the chemical tional ﬂoat glass (T is at around 980 C, therefore compositions derived by the XRF analyses, the APF 80–100 C lower than for SLS). It contains K O and and G are calculated and listed in Table 9. B O in small amounts (< 3 wt%), and has a higher 2 3 Therefore, by reviewing Table 9, it is anticipated Na O/CaO ratio than typical SLS recipes. Despite the that the lead silicate glass samples, which present the 2 slightly lower E modulus than the one of AGC dark lowest dissociation energy and packing density, will blue glass, it presented the highest ﬂexural strength have the lowest strength as well, while the soda lime among all tested specimens. This is attributed to a lower brittleness of this particular glass. Sehgal and Ito (1998) state that a higher molar volume (V ) plays a key role The atomic packing density is calculated using the following in the reduction of the brittleness, as a more open struc- formulas: ture allows more deformation prior to crack initiation. More speciﬁcally, an increasing soda/calcia ratio would x · V i i APF = (9) decrease the brittleness, as well as the partial substitu- tion of soda for potash. This is in accordance with the compositional variations of Poesia glass to the typical where x is the molar fraction, V the ionic volume of the ith i i SLS recipe, which contribute to a more open structure oxide and V is the molar volume of glass, and speciﬁcally: (Fig. 35) that allows for a slightly increased accom- modation of the stresses around the point/ﬂaw where 4 ◦ 3 3 the crack will initiate. The “Poesia 1070 C specimens V = · π · N · (x · r + y · r ) (10) i A A B failed due to machining damage (Fig. 36). The Wertheim glass has the highest measured and Young’s modulus and the highest calculated total dis- sociation energy, while it’s calculated molar volume is similar to the Poesia glass. The higher stiffness (in com- V = (11) parison to an SLS glass) can be attributed to the partial substitution of silica with alumina (≈ 5%), that reduces the openness of the network Sehgal and Ito (1998). Sim- where N is the Avogadro’s number, r = ionic radii of M O A A,B x y oxide, M is the molecular mass and ρ is the density of the glass. ilar to the Poesia glass, it has a lower forming temper- The V is derived from Makishima and Mackenzie (1973)and ◦ ature than SLS glasses (T at around 1015 C), which Inaba et al. (1999) based on Pauling’s ionic radii. The density is calculated from the chemical composition using the model developed by Fluegel (2007b). The total dissociation energy is calculated from the dissocia- 23 It should be noted that the calculated E using the APF Poesia tion energy of the oxide constituents listed in the work of Inaba and G corresponding to the chemical composition is found much et al. (1999). lower than the E . This could be related to a wrong esti- AGC blue PbO has one of the lowest Gi as reported by Makishima and mate of the B O content, which cannot be determined by the 2 3 Mackenzie (1973), and a relatively high molar atomic mass. The XRF analysis, and/or a higher packing density attributable to the increased mass of the lead ion slows down the chemical reac- thermal history of the kiln-cast components. The E derived Poesia tions during quenching, and results in a less organized/packed from the LVDT data is thus considered more reliable for further network. analysis. 123 Investigating the ﬂexural strength of recycled cast glass 477 Fig. 35 Graph of total dissociation energy versus the molar volume Fig. 36 a Microscope image of the bottom surface and corner, the fracture mirror elongation to the left and the consecutive and the fracture surface of a “Poesia 1070 C” specimen. The machining crack hackles along the fractured edge. b close-up of cause of failure is grinding damage, which is demonstrated by the fracture origin can be linked to the mixed alkali effect and the pres- reason this glass failed at a lower stress is linked to ◦ ◦ ence of boron trioxide in a small quantity (Morey 1932). its kiln-casting at temperatures (820 C, 900 C) well According to the E/V properties of this glass, a much below its liquidus point, which resulted in evident inho- higher ﬂexural strength should have been expected. The mogeneities. These inhomogeneities are concentrated in the interface created between each pellet of glass, The term describes anomalies observed in glasses and melts and compose a 3-dimensional network of planar zones containing a mixture of two or more alkali oxides. According to consisting of bubbles and loose crystal formations. In Shelby (2005), the viscosities of such melts are lower than those addition, the forming temperature favours the occur- containing the same amount of a single alkali oxide. 123 478 T. Bristogianni et al. Fig. 37 Fracture surface of the “Wertheim 900 C” specimen. ror and the crystalline interface located just below the fractured a Microscope image showing the bottom surface and fracture surface (only the crystal formations along the bottom surface are mirror. The specimen failed from a grinding scratch (red arrow) fractured) next to the fusion interface (white arrows). b Close-up of the mir- rence of stones, due to the reaction of the hot glass will heat up faster. In a similar manner, the dark glass with the mould, which are sufﬁciently sub-surface that will set faster during cooling due to the greater heat they cannot be entirely removed during standard post- loss by radiation (Kita˘ıgorodski˘ı and Solomin 1934; processing. These stones seem to weaken the glass sur- Burch and Babcock 1938). The faster setting rate can face and contribute to the formation of deeper striations inﬂuence the coordination state of the transition metal during grinding, which are the sources of failure. The oxides included in the composition and thus affect above described 3D network may not be responsible for the total dissociation energy of the network bonds— the crack origin, but given that the specimens fail from something not accounted for in the calculations. In a ﬂaw in close proximity to the network, it may con- addition, the lower liquidus point and increased heat tribute to zones of concentrated stress along the surface absorption promote the full fusion of the cullet pieces (Fig. 37). and the elimination of stone formation, thus leading to It is also interesting to compare the “AGC dark blue the diminishing of ﬂaws at the glass surface, and to ﬂoat” to the “FT ﬂoat” specimens. These two glasses a higher ﬂexural strength. In antithesis, infolds at the have very similar compositions and are almost identical glass surface of the “FT ﬂoat” specimens are created by in calculated atomic packing density and total dissoci- insufﬁcient fusion of the cullet positioned next to the ation energy. However, the measured Young’s modulus mould walls, and crystalline inclusions due to contam- of the “FT ﬂoat” glass specimens is lower, and so is the ination from the mould, are the main cause of failure ﬂexural strength. This is probably linked to the thermal of the “FT ﬂoat” samples, according to the analysis of history of these two glasses. On the one hand, the “AGC the fracture mirrors (Fig. 38). Due to these ﬂaws, the dark blue” glass has a slightly lower liquidus point (T “FT ﬂoat” glass specimens fail at values lower than ◦ ◦ is around 1046 C, while for the “FT Float” is 1063 C). expected in comparison to the rest of the samples. On the other hand, the dark colour of the AGC glass seems to contribute to the quality of the casting. The The XRF identiﬁes a series of transition metals in this glass dense dark blue colour absorbs more infra-red light dur- that act as colorants: 0.76% Fe O , 0.065% TiO , 0.029% MnO, 2 3 2 ing heating than the transparent light blue, thus the body 0.023% Cr O ). 2 3 123 Investigating the ﬂexural strength of recycled cast glass 479 Fig. 38 Microscope images of the bottom and fracture surface The hertzian cones (grey arrows) that refer to impact damage of a “FT Float 1120 C” specimen, showing the fracture origin have an opposite to the crack front direction and are considered (Fig. 38b is a magniﬁcation of the fracture origin). The cause of secondary breaks. Looking through the hertzian cones, traces of failure is a combination of grinding scratches (red arrows) acting crystalline inclusions can be observed upon a surface infold with crystalline inclusions (white arrows). Regarding the fracture analysis of the glasses stud- specimens described above. All of the studied spec- ied in this category, the most prevailing causes of fail- imens, “Float combo”, “Oven doors”, “Car wind- ure are found to be machining damage and handling shields”, and “AGC enamel black” are typical soda lime ﬂaws (see also Fig. 28, most “pure” specimens fail at silicates and have a large amount of distinct crystalline an edge ﬂaw), justifying that the purity of the cullet and inclusions, and/or heavy cord. Their ﬂexural strength the relatively high forming temperatures (in compari- is slightly lower than the one observed for the FT Float son with other glass samples in this work) eliminate specimens and their Young’s moduli are comparable. the quantity of strength limiting ﬂaws. Exceptions are The “AGC enamel black” series seems to have the high- found in the “FT ﬂoat” series, as described above, and est ﬂexural strength in this category, which is attributed the “Schott DURAN tubes” specimens. These borosil- to the fact that only one type of glass is used for the cast- icate glass samples are in fact formed at a high viscos- ing of these samples (thus no cord is observed due to 4.5 ity (≈ 10 dPa s < T ) and are characterized by an minor compositional variations). In addition, the sub- increased amount of bubbles (mainly concentrated at stantial size of the glass pieces allows their thorough the interfaces created between each cullet piece dur- cleaning, which is not the case in the smaller sized cul- ing forming). These bubbles form clusters for crystal let of the “Oven door” and “Car windshield” samples. growth and if located at the glass surface or at close All specimens fail at lower values than most of the proximity, they become the strength limiting ﬂaw that purer glasses studied above, mainly due to crystalline leads to fracture (Fig. 39). The ﬂexural data obtained formations at the surface (Fig. 40). These stones are for these two glasses from the 1st series of four-point created either from inherent contamination, or from bending experiments match with the results of the sec- further reaction of the contaminants with the mould ond test. material. The multiple defects located in the bulk of these specimens are not activated during the 4-point (b) Contaminated vs. non-contaminated glass speci- bending nor do they seem to reduce the Young’s mod- mens ulus. On the contrary these defects are tolerated within the glass network. However, the more the defects in the This category studies glass specimens kiln-cast from bulk, the more the chances of such ﬂaws to be exposed contaminated cullet, and compares them to the purer 123 480 T. Bristogianni et al. Fig. 39 Microscope images of the fracture origin of a “Schott not break at a ﬂaw exactly at the surface. b Magniﬁcation of the DURAN tubes 1120 C” specimen. a Incomplete fracture mirror crystalline inclusion, which is clustered together with air bub- around the origin ﬂaw, which is a crystalline inclusion at 1 mm bles. Early mist and hackle appear within the mirror as a result inside from the bottom surface. This is the only specimen that did of interaction of the elastic wave with the defect Fig. 40 Microscope images of crystalline formations that func- seem to promote such formations. a The stone in the “AGC Float tion as the origin of fracture in contaminated kiln-cast specimens. with black enamel, 1120 C” specimen is adjacent to the enamel The reaction of cullet contaminants (e.g. coatings) with the mould interface. b Stone in a “Oven doors 1120 C” specimen ◦ ◦ ◦ at the surface, and consequently the higher the risk of 1070 C and/or 1120 C (“FT Float 1120 C”, “Schott ◦ ◦ failure. DURAN 10 vertical layers 1070 C, 1120 C”). The particular aspect with this category is that the “defects” (c) Non-contaminated homogeneous glass specimens or zones of compositional/structural variation, are vs. with crystallized interfaces deliberately engineered at speciﬁed locations and geo- In this category, the soda lime and borosilicate glass metrical patterns. Thus, in antithesis with the random occurrence of stones described in the category above, in samples that are kiln-formed at 970 C and contain structured crystallized interfaces (“Float 1cm, 3 hori- this section, the size and distribution of the crystalline formations can be anticipated. As a consequence, their zontal layers”, “Float 1cm, 24 vertical layers”, “Schott DURAN 10 vertical layers”), are studied in comparison effect on the structural performance can be directly cor- related. to their more homogeneous versions, kiln-formed at 123 Investigating the ﬂexural strength of recycled cast glass 481 Fig. 41 Fracture pattern and ﬂexural strength range (MPa) of lized interface is exposed at the bottom surface, the specimens homogeneous (top) and with crystallized interfaces (bottom) fail at a lower force, from a ﬂaw originating at the glass-crystal specimens. Note that if the crystallized interface is situated only interface. Especially in the case of the borosilicate crystallized in the bulk (bottom right), the bending strength of the specimen specimens, the low elastic energy stored results in a single crack is similar to a homogeneous one (top right). When the crystal- without forking Therefore, it can be observed that the fused “Float tion should be raised to the intermediate states between 1cm, 3 horizontal layers” specimens present very sim- a fused glass specimen produced at viscosities around ilar ﬂexural strength and Young’s modulus with the 10 dPa s and a homogeneous specimen cast at temper- more homogeneous “FT Float” specimens (Fig. 41). atures well above the liquidus point (where the rate of This is because the crystalline interfaces are located diffusion is much higher). Specimens produced at a 10 in the bulk, in parallel layers to the bottom surface, or even a 10 dPa s viscosity seem to retain traces of the and thus are not exposed to the maximum tensile stress interface created between each cullet piece during heat- zone. They behave therefore in a similar manner to ing up, in the form of subtle bubble veils, cord and spots the homogeneous specimens. This is not the case how- of crystalline formation. This is evident for example in ever with the “Float 1cm, 24 vertical layers” specimens, the “Schott DURAN 10 vertical layers 1070 C”, kiln- where the crystalline interfaces are exposed at the bot- cast at a 10 dPa s viscosity (Fig. 42). These samples tom surface, and in fact aligned perpendicularly to the contained the above described bubble veils and stones tensile forces. Although the Young’s modulus remains in the same geometrical arrangement as the “Schott similar, the ﬂexural strength is reduced by more than DURAN 10 vertical layers 970 C”. These specimens, 20%. The fracture origin of these samples is always although stronger than the fused version, had a 30% located at these crystal-glass interfaces and initiates lower ﬂexural strength than the specimens kiln-cast at from the glass zone in immediate proximity. The crys- a50 C higher temperature. They all failed from either talline formations thus seem to act as stress inducing a crystalline ﬂaw or a bubble located in one of these elements, of perhaps higher fracture toughness than the veils (Fig. 43). This is a very crucial issue, given the fact ◦ ◦ surrounding glass matrix, which weaken the glass spec- that both the 1070 C and 1120 C produced specimens imen. look the same and are transparent and not comparable The type and thickness of the crystalline interface to the contaminated samples described in the category also plays a signiﬁcant role. The thin b-cristobalite above. This highlights how critical a 50 C tempera- layer created in “Schott DURAN 10 vertical layers ture difference can be when casting at viscosities just 970 C” results in a dramatic drop of 75% of the around and below the liquidus point. strength, and a decrease of the Young’s modulus. The fracture origin is, in a similar manner to the ﬂoat exam- (d) Reference specimens ples, always located at the crystalline-glass interface. At this point, knowing the effect of these crystalline The industrially manufactured glass specimens are formations and their geometrical arrangement, atten- tested in order to provide a point of reference and com- 123 482 T. Bristogianni et al. Fig. 42 “Schott DURAN 10 vertical layers” specimens pro- perature result in a lower ﬂexural strength. The origin of fracture ◦ ◦ duced at 1070 C (left column) and 1120 C (right column). The in these specimens can be found in one of these bubble veils remnant bubble veils in the specimens produced at a lower tem- Fig. 43 Microscope images of the fracture mirror of a “Schott Their perpendicular to the surface direction and the presence of DURAN 10 vertical layers, 1070 C” specimen. a Succession of a bubble, suggest that these stones are formed from the interac- air bubbles close to the bottom surface, interacting with the elas- tion of the mould material with the bubbles created at the fusion tic wave. The cause of fracture is a stone formation (right end of interface between the glass rods during forming at a-favourable the picture) in proximity to the bubble clustering. b Magniﬁca- for crystallization-temperature tion of stone formations extending from the surface to the bulk. parison with the kiln-cast glass samples. Their struc- faster annealing scheme followed for these compo- tural performance is described below per type. nents, which causes residual stresses frozen in the glass, The beams cut out from standard Poesia cast glass and makes it more susceptible to damage. As a conse- bricks (originally hot poured at around 1200 C) are quence, during the cutting and grinding of the com- more homogeneous than the kiln-cast specimens pro- ponent in size, multiple chips and resulting cleavage duced in the lab. Apart from some minor striae and damage are caused due to insufﬁcient annealing, which few bubbles, they do not contain critical defects such are not entirely removed during polishing. The added as stones, since the purity of the raw source, the stress and machining defects are the cause of fracture, above liquidus point forming temperature, the abrupt at a lower strength. quenching at atmospheric conditions, and the stainless Considering the single ﬂoat glass pane specimens, steel moulds used for their casting, prevent their for- these are the most homogeneous of all studied sam- mation. Nonetheless, these specimens present a 10% ples, with a pristine polished bottom and top surface. lower ﬂexural strength than the less homogeneous, re- Since these specimens are cut out from larger ﬂoat pan- cast specimens at 1070 C. This is attributed to the els, their edges are ground and polished as described in 123 Investigating the ﬂexural strength of recycled cast glass 483 Sect. 2.1. All single pane samples from the ﬁrst series of at relatively low temperatures (820–1120 C), and the four-point bending experiments failed from a machin- ﬂexural strength of the kiln-cast specimens is evaluated. ing ﬂaw at the edge, within the bottom zone of maxi- The kiln-casting experiments show that meticulous mum tensile stress. The average ﬂexural strength for the separation of cullet at the recycling facilities guaran- 10 mm panes is 55 MPa, which is 20% higher than the tees a successful casting. Coatings and traces of exter- “FT Float 1120 C” specimens but 20% lower than the nal contaminants such as organics and metals are tol- highest scoring specimens “AGC dark blue 1120 C” erated by the glass network yet lead to defects and low and the “Poesia 1070 C”. Undoubtedly, the quality of ﬂexural strength, while contamination by glass ceram- the bottom edges can dramatically affect the ﬂexural ics and glasses with signiﬁcant compositional varia- strength of the ﬂoat glass sample in bending. Accord- tions causes the fracture of the specimens during cool- ing to the size and the polishing quality of the samples, ing. Glass compositions with a lower liquidus point and the test settings, a wide range of ﬂexural strengths facilitate low temperature kiln-casting which leads to in 4 point-bending are reported in literature, from 35 more homogeneous glass surfaces, as the lower viscos- to 170 MPa (Veer and Rodichev 2011), 51–71.5 MPa ity during forming minimizes the occurrence of sinter- (Veer 2007), 53–129 MPa (Yankelevsky et al. 2016)to ing ﬂaws, surfaces bubbles and stone formation from name a few. The 55 MPa strength of the tested samples mould contamination. in this study is at the low end of this range, and in line Regarding the four-point bending experiments, with the literature, given the relatively rough edge ﬁn- although the number of tested specimens per glass type ishes. A much higher strength could be expected with is not sufﬁcient for deriving statistical data, they do pro- ﬁner polishing. In that sense and taking into account vide a good overview and reasonable estimate of the that the kiln-cast specimens exhibiting higher tensile structural performance of each specimen type, accord- strengths failed as well from machining ﬂaws, it can be ing to the chemical composition, level of contamina- derived that a much higher strength is possible with the tion, and followed casting parameters. industrial ﬁne polishing of the kiln-cast specimens. The effect of the chemical composition on the The beams produced from adhesively bonded (Delo strength is distinctly observed in the specimens pro- Photobond 4468) 8/10 mm thick ﬂoat glass plies, and duced from purer cullet and at higher forming temper- tested with their plies parallel to the bottom surface, atures. Among these samples, a clear increase in the have an average ﬂexural strength of 48 MPa (1st and strength and Young’s modulus is observed, consecu- 2nd four point bending series), which is 10% higher tively from the lead silicate, to the borosilicate, barium than the kiln-formed “FT Float 1120 C” specimens. silicate and up to the soda lime silicate family. The None of the specimens failed from an edge ﬂaw; the purer, more homogeneous samples predominantly fail cause of failure is attributed to minor handling dam- from external defects induced by machining and han- age at the bottom surface. The Young’s modulus of dling damage. The effect of the composition is however the adhesively bonded beams is lower than that of the blurred in the more contaminated samples, where crys- monolithic, kiln-cast SLS specimens, due to the adhe- talline formations formed at the bottom surface within sive layers. the zone of maximum tensile stress, are the prevail- Overall, the ﬂexural strength values obtained from ing cause of fracture leading to a signiﬁcantly lower the industrially produced reference samples are at the strength. top end of the 30–55 MPa (second) zone, and do not Within the soda lime silica family, particularly exceed the performance of the purest kiln-cast samples promising are the slightly modiﬁed recipes containing (found in the ﬁrst zone). This is an encouraging result, small amounts of K O and B O and a higher Na O 2 2 3 2 given the fact that all the kiln-cast specimens produced to CaO ratio. The lower viscosity of these glass melts for this study have some level of inhomogeneities. facilitates the casting process, while their more open structure (higher molar volume) presents a less brittle alternative for a similar Young’s modulus to that of SLS 5 Conclusions glasses, leading eventually to a higher ﬂexural strength. Glass families of an even lower liquidus point, A variety of commercial glass waste types is tested for such as the studied lead silicate and barium silicate the ability to be kiln-cast into structural components samples, are attractive for lower energy manufactur- 123 484 T. Bristogianni et al. ing. However, for structural applications demanding ﬂoat glass kiln-cast specimens (at 1120 C), yet score higher strength, the barium silicate option is much more at the lower end of strength values reported in the liter- promising due to the higher E modulus and less sus- ature. Machining ﬂaws from the processing to size, and ceptibility to scratching. insufﬁcient annealing in the case of the cast bricks, are Regarding the more inhomogeneous specimens, the factors responsible for the lower strength. A ﬁner produced from contaminated cullet at temperatures polishing would signiﬁcantly increase the strength, not around the liquidus point, they still present a good ﬂex- only of these samples, but also of the purer kiln-cast ural strength and are suitable for structural applications specimens. However, given that the lowest strength demanding lower tensile strength, such as bricks. The specimens would be less affected by a ﬁner polishing ﬂaws occurring in the bulk are not activated during the quality, the statistical strength would not be increased four-point bending test and have a minor or even negli- that much, as it is dominated by these lower outliers. gible contribution to the strength and E modulus. How- ever, an increased density of defects in the bulk should imply a higher density of ﬂaws at the surface as well, 6 Recommendations which should lead to an average strength reduction. A higher forming temperature (above the liquidus point) The results of this study show the potential of recycling would signiﬁcantly help in diminishing the amount of waste glass into cast structural building components. ﬂaws, but considering the economic and environmen- However, for the safe application of such products, tal advantages of lower temperature processing, such an further validation is required and an increased number act would be only meaningful if higher design strengths of tested specimens per category (≥ 30, Quinn et al. were required per speciﬁc case. 2009) is needed to derive statistical predictions. In par- Crystallized geometrical structures are induced ticular, the repetition of testing is of crucial importance within soda-lime-silica and borosilicate specimens pro- in the case of the contaminated samples, where a higher 6 5 duced at higher viscosities (10 –10 dPa s). If these degree of variability is expected in the mechanical prop- structures are located in the bulk, the ﬂexural strength erties. The systematic testing of such samples should of the specimen is equal to that of a more homoge- be linked with a quantiﬁed documentation of the type neous casting at a higher temperature (close to the and level of inhomogeneities in the glass prior to test- liquidus point). However, the exposure of such struc- ing. Careful and extensive fracture analysis of the tested tures at the surface subjected in tension can lead to specimens is also necessary to identify the most critical a dramatic decrease of strength of even 75% accord- defects, and the relationship of the ﬂaw size to the ﬂex- ing to the nature of the produced crystalline forma- ural strength. The physiochemical identiﬁcation of the tions. In this case, the origin of fracture always occurs crystalline formations at the glass surface by scanning in the glass/crystal interface. Speciﬁc attention should electron microscopy is required for categorizing such 5 4 be given to castings formed at between 10 –10 dPa s critical ﬂaws. Further testing is necessary, as well, to viscosities, as the glass products may appear homoge- determine the inﬂuence of scale factor, and of static neous but retain signiﬁcant inhomogeneous zones of fatigue in moist environments (effect of slow crack miniscule bubbles and stones at the former interface growth). In addition, the studying of the behaviour of created between each cullet pieces during heating up. crystalline inclusions in the bulk glass under thermal Such formations exposed at the tensile surface are crit- gradients relevant to building applications is important ical for the specimen’s strength. to eliminate the risk of thermal cracking. Additional, Industrially SLS manufactured glass samples, post- non-destructive testing for determining the Young’s processed in the lab facilities to match the studied spec- modulus and the level of inhomogeneities in the cast imen size, present similar ﬂexural strength to that of the glass is also suggested, implementing the Impulse Exci- tation Technique. Investigation of whether such a fast Yamane and Mackenzie (1974) prove in their model the pro- and inexpensive non-destructive technique could serve portional relationship of Vicker’s hardness to the Young’s mod- as a quality control method for cast glass products is ulus and bond strength. As a point of reference, Ainsworth’s worth exploring. (1954) measurement of Vicker’s hardness for a 18Na 0 ·10BaO · 2 Regarding the more contaminated components, 72SiO (mol%) glass is 522 kg/mm and for a 18Na 0 · 10PbO · 2 2 72SiO (mol%) glass 445 kg/mm . attention should be given in improving the quality of the 123 Investigating the ﬂexural strength of recycled cast glass 485 Fig. 44 Prototypes of composite kiln-cast glass components, tallized zone produced from waste glass powder (residue from containing a purer glass at the bottom and a weaker, more con- the glass container recycling process). b Reinforcement of a “FT taminated glass at the bulk and top zone. a Gradient from a pure Float” glass beam by a bottom layer comprising the stronger transparent soda lime silica glass (bottom) to a partially crys- “ACG blue” glass stone-containing surface of the recycled glass. Chemi- also very grateful to Cor Wittekoek (Vlakglasrecycling), Danny Timmers (Maltha Glasrecycling Nederland), Marco Zaccaria and cal strengthening of the surface by ion exchange could François Boland (AGC Belgium), Brian Wittekoek (Coolrec), be a—high cost—solution although this is not likely to Bettina Sommer (Royal Leerdam Crystal), and Klaas Roelf- help with deep defects. Another simpler solution appli- sema (Schott), for the glass cullet contribution, which was of key cable for high viscosity castings (where the diffusion importance for this work. Finally, we would like to thank Erik Muijsenberg (Glass Service) for the “Glass Defects” (Bartuška rate is low), is the structuring of two (compatible) cullet 2008) book, and Peter de Haan (AGR Delft) for the “Color Atlas qualities inside the mould: a purer along the demand- of Container Defects” (Aldinger and de Haan 2019) and “Color ing zones, and a lower more contaminated quality in the Atlas of Stones in Glass” (Aldinger and Collins 2016) books, bulk (Fig. 44). Such a composite glass would enable the which were very helpful in the process of defect categorization and identiﬁcation. use of contaminated, unwanted cullet without necessar- ily compromising the strength of the ﬁnal product. Compliance with ethical standards Lastly, the engineering of crystalline or bubble veil geometrical structures within the glass is worth further Conﬂict of interest On behalf of all authors, the corresponding exploration, as they can lead to fractures within a pre- author states that there is no conﬂict of interest. dictable strength range and location. They also can lead to building components with non-standard appearance Open Access This article is licensed under a Creative Com- and thus higher architectural appeal. mons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original Acknowledgements The authors would like to express their author(s) and the source, provide a link to the Creative Com- gratitude to Giorgos Stamoulis for his signiﬁcant contribution mons licence, and indicate if changes were made. The images or in preparing the four-point bending experiment and DIC mea- other third party material in this article are included in the article’s surements, as well as Kees van Beek for his guidance. We are Creative Commons licence, unless indicated otherwise in a credit particularly grateful to George Quinn and James Varner for their line to the material. If material is not included in the article’s Cre- valuable input regarding the experimental procedure and the frac- ative Commons licence and your intended use is not permitted by ture analysis of the results. The feedback received from Bert statutory regulation or exceeds the permitted use, you will need Sluijs, Mauro Overend, Christian Louter and Karl-Heinz Wolf to obtain permission directly from the copyright holder. To view is highly appreciated. We would also like to thank Ruud Hen- a copy of this licence, visit http://creativecommons.org/licenses/ drikx for the XRF and XRD analyses, Mariska van der Velden by/4.0/. for assisting in the specimen preparation, Sander van Asperen and Wolfgang Gard for providing access to their labs’ Keyence digital microscopes, and Henning Katte and Daniel Schreinert (Ilis) for sponsoring the use of StrainScope Flex. The authors are 123 486 T. Bristogianni et al. References Heimerl, W.: Chemical resistance and corrosion, and ion release. In: Bach, H., Krause, D. (eds.) Analysis of the Composition and Structure of Glass and Glass Ceramics. Springer, Berlin Abrisa Technologies: SCHOTT Duran Lab Glass (Tubed) (1999) (2014) Höland, W., Beall, G.H.: Glass—Ceramic Technology, 3rd edn. Ainsworth, L.: The diamond pyramid hardness of glass in relation Wiley, Hoboken (2020) to the strength and structure of glass. Soc. Glass Technol. Inaba, S., Fujino, S., Morinaga, K.: Young’s modulus and compo- 38, 501–547 (1954) sitional parameters of oxide glasses. J. Am. Ceram. Soc. 82, Aldinger, B.S., Collins, B.K.: Color Atlas of Stones in Glass. 3501–3507 (1999). https://doi.org/10.1111/j.1151-2916. American Glass Research, Butler (2016) 1999.tb02272.x Aldinger, B.S., de Haan, P.W.: Color Atlas of Glass Container Ito, S., Taniguchi, T.: Effect of cooling rate on structure and Defects. American Glass Research, Butler (2019) mechanical behavior of glass by MD simulation. J. Non- Andreola, F., Barbieri, L., Corradi, A., Lancellotti, I., Falcone, Cryst. Solids 349, 173–179 (2004). https://doi.org/10.1016/ R., Hreglich, S.: Glass-ceramics obtained by the recycling of j.jnoncrysol.2004.08.180 end of life cathode ray tubes glasses. Waste Manag. 25, 183– Kita˘ıgorodski˘ı, I.I., Solomin, N.W.: Rate of setting of glass dur- 189 (2005). https://doi.org/10.1016/j.wasman.2004.12.007 ing working. Soc. Glass Technol.J. 18, 323–335 (1934) ASM International: Applications for Glasses Engineered Mate- Makishima, A., Mackenzie, J.D.: Direct calculation of Young’s rials Handbook Desk Edition. ASM International (1995) moidulus of glass. J. Non-Cryst. Solids 12(1), 35–45 (1973). Bartuška, M.: Glass Defects. Glass Service Inc. and Práh, Prague https://doi.org/10.1016/0022-3093(73)90053-7 (2008) Martienssen, W., Warlimont, H.: Springer Handbook of Con- Brennan, J.J.: Program to Study SiC Fiber Reinforced Glass densed Matter and Materials Data (2005) Matrix Composites. United Technologies Research Center, MatWeb: C-Glass Fiber. www.matweb.com Connecticut (1979) Montazerian, M., Singh, S.P., Zanotto, E.: An analysis of glass- Bristogianni, T., Oikonomopoulou, F., Justino de Lima, C.L., ceramic research and commercialization. Am. Ceram. Soc. Veer, F.A., Nijsse, R.: Structural Cast Glass Components Bull. 94, 30–35 (2015) Manufactured from Waste Glass: Diverting Everyday Dis- Morey, G.W.: The effect of boric oxide on the devitriﬁca- carded Glass from the Landﬁll to the Building Industry. tion of the soda-lime-silica glass Na O–CaO–B O –SiO . 2 2 3 21 Heron 63 (1/2 Special issue: Structural Glass) (2018) Quat. Syst. 15(9), 457–475 (1932). https://doi.org/10.1111/ Bristogianni, T., Oikonomopoulou, F., Veer, F.A., Nijsse, R.: The j.1151-2916.1932.tb13959.x effect of manufacturing ﬂaws in the meso-structure of cast Mueller, J., Boehm, M., Drummond, C.: Direction of CRT waste glass on the structural performance. In: Zingoni, A. (ed.) glass processing: electronics recycling industry communi- Advances in Engineering Materials, Structures and Sys- cation. Waste Manag. 32, 1560–1565 (2012). https://doi. tems: Innovations, Mechanics and Applications, pp. 1703– org/10.1016/j.wasman.2012.03.004 1708. CRC Press, Leiden (2019) National Institutes of Health (NIH): PubChem database. https:// Burch, O.G., Babcock, C.L.: Effect of glass color on setting rates pubchem.ncbi.nlm.nih.gov/ in manufacture of glass bottles. J. Am. Ceram. Soc. 21(10), Oikonomopoulou, F., Bristogianni, T., Veer, F.A., Nijsse, R.: 345–351 (1938) The construction of the crystal Houses façade: challenges Campbell, D.E., Hagy, H. E.: Glasses and Glass-Ceramics. In: and innovations. Glass Struct. Eng. 3(1), 87–108 (2018a). Lynch, C.T. (Ed.) CRC Handbook of Materials Science, https://doi.org/10.1007/s40940-017-0039-4 vol. II: Material Composites and Refractory Materials. CRC Oikonomopoulou, F., Bristogianni, T., Barou, L., Jacobs, E., Press, USA (1975) Frigo, G., Veer, F.A., Nijsse, R.: Interlocking cast glass com- Chyung, K.: Transparent Beta-Quartz Glass Ceramics. Corning ponents, exploring a demountable dry-assembly structural Glass Works, United States Patent US4018612A (1977) glass system. Heron 63, 103–138 (2018b) Corning: Properties of Corning’s Glass and Glass Ceramic Fam- Oikonomopoulou, F., Bristogianni, T., Barou, L., Veer, F.A., ilies. In: Materials for the Design Engineer. USA (1979) Nijsse, R.: The potential of cast glass in structural applica- Fluegel, A.: Glass viscosity calculation based on a global statis- tions. Lessons learned from large-scale castings and state- tical modeling approach. Glass Technol. Eur. J. Glass Sci. of-the art load-bearing cast glass in architecture. J. Build. Technol. A 48, 13–30 (2007a) Eng. 20, 213–234 (2018c). https://doi.org/10.1016/j.jobe. Fluegel, A.: Global model for calculating room-temperature glass 2018.07.014 density from the composition. J. Am. Ceram. Soc. 90, Priven, A.I.: Evaluation of the fraction of fourfold-coordinated 2622–2625 (2007b). https://doi.org/10.1111/j.1551-2916. boron in oxide glasses from their composition. Glass 2007.01751.x Phys. Chem. 26(5), 441–454 (2000). https://doi.org/10. Friedrich & Dimmock Inc.: Comparative Values of Borosilicate 1007/BF02732065 Glasses. In: Simax Glass Properties Quinn, G.D.: NIST Recommended Practice Guide: Fractography Gold Star: Investment Casting Powder Safety Data Sheet, www. of Ceramics and Glasses, 2nd Ed. (2016) goldstarpowders.com (2019) Quinn, G.D., Swab, J.J.: Fracture toughness of glasses as mea- Goodwin Refractory Services Ltd: Crystalcast (M248). www. sured by the SCF and SEPB methods. J. Eur. Ceram. grscastingpowders.com (2003) Soc. 37(14), 4243–4257 (2017). https://doi.org/10.1016/j. Gregory, C.: Hollow Fibers. In: Sanghera, J.S.A., I.D. (ed.) jeurceramsoc.2017.05.012 Infrared Fiber Optics. CRC Press LLC, USA (1998) Quinn, G.D., Ives, L.K., Jahanmir, S.: On the nature of machining cracks in ground ceramics: part II—comparison to other 123 Investigating the ﬂexural strength of recycled cast glass 487 silicon nitrides and damage maps. Mach. Sci. Technol. 9, Veer, F.A., Rodichev, Y.: The structural strength of glass: hidden 211–237 (2005). https://doi.org/10.1081/MST-200059051 damage. Strength Mater. 43, 302–315 (2011). https://doi. Quinn, G.D., Sparenberg, B.T., Koshy, P., Ives, L.K., Jahanmir, org/10.1007/s11223-011-9298-5 S., Arola, D.D.: Flexural strength of ceramic and glass rods. Volf, M.B.: Chemical Approach to Glass. Elsevier, Amsterdam J. Test. Eval. 37, 222–244 (2009). https://doi.org/10.1520/ (1984) JTE101649 Yamane, M., Mackenzie, J.D.: Vicker’s Hardness of glass. J. Schott: Tubular Glass Photobioreactors (2015a) Non-Cryst. Solids 15(2), 153–164 (1974). https://doi.org/ Schott: NEXTREMA (2015b) 10.1016/0022-3093(74)90044-1 Schott: DURAN Technical Data (2017) Yankelevsky, D., Spiller, K., Packer, J., Seica, M.: Fracture Sehgal, J., Ito, S.: A New Low-Brittleness Glass in the Soda- characteristics of laboratory-tested soda lime glass spec- Lime-Silica Glass Family. 81(9), 2485–2488 (1998). https:// imens. Can. J. Civ. Eng. (2016). https://doi.org/10.1139/ doi.org/10.1111/j.1151-2916.1998.tb02649.x cjce-2016-0374 Shelby, J.E.: Introduction to Glass Science and Technology. The Yun, Y.H., Bray, P.J.: Nuclear magnetic resonance studies of Royal Society of Chemistry, UK (2005) the glasses in the system Na O–B O –SiO .J.Non- 2 2 3 2 Silva, R.V., de Brito, J., Lye, C.Q., Dhir, R.K.: The role of glass Cryst. Solids 27(3), 363–380 (1978). https://doi.org/10. waste in the production of ceramic-based products and other 1016/0022-3093(78)90020-0 applications: a review. J. Clean. Prod. 167, 346–364 (2017). Zhdanov, S.P., Shmidel’, G.: Coordination state of boron in https://doi.org/10.1016/j.jclepro.2017.08.185 sodium borosilicate glasses from NMR data. Fiz. Khim. textcircled Songhan Plastic Technology Co., L.: Schott Nextrema Stekla 1(5), 452–456 (1975) 724-3, 712-3 Glass Ceramic TM Specialty Glass Products: STARPHIRE Ultra-Clear Soda Publisher’s Note Springer Nature remains neutral with regard Lime Glass to jurisdictional claims in published maps and institutional afﬁl- Thompson, D.A.: Low liquidus glasses for television tube face- iations. plates. US Patent US4331770A (1980) Veer, F.A.: The strength of glass, a nontransparent value. Heron 52, 87–104 (2007)
Glass Structures & Engineering – Springer Journals
Published: Nov 2, 2020
Access the full text.
Sign up today, get DeepDyve free for 14 days.