The Use of Tailings to Make Glass as an Alternative for Sustainable Environmental Remediation: The Case of Osor, Catalonia, Spain

Tailings from the Osor fluorite mines release large amounts of potentially toxic elements into the environment. This work is a proposal to remove these waste materials and use them as a raw material in the manufacture of glass. The chemical composition of the tailings was determined by X-ray fluorescence and the mineralogy by X-ray diffraction. Waste materials have SiO2, Al2O3 and CaO contents suitable for a glass production, but Na as NaCO3 has to be added. Two glass formulations, with 80–90% of the residue and 10–20% Na2CO3, have been produced. The crystallization temperatures, obtained by differential thermal analysis, were 875 and 901 ◦C, and the melting temperatures were 1220 and 1215 ◦C for the G80-20 and G90-10 glasses, respectively. The transition temperatures of glass were 637 and 628 ◦C. The crystalline phases formed in the thermal treatment to produce devitrification were nepheline, plagioclase and diopside in the G80-20 glass, and plagioclase and akermanite-gehlenite in the G90-10 glass. The temperatures for the fixed viscosity points, the working temperatures and the coefficient of expansion were obtained. The chemical stability of the glass was tested and results indicate that the potentially toxic elements of the tailings were incorporated into the glass structure.


Introduction
Tailings produce a large number of environmental impacts [1]. Especially dangerous are the problems that occur when the tailings result from the processing of sulphide ores. In these cases the interaction of meteoric water with the sulphides remaining in the tailings drives the generation of acid drainage, which can carry large amounts of potentially toxic elements [2][3][4]. These elements are liberated into the environment and produce pollution in water, which cannot be used for other applications. Special care must be taken in those tailings with small particles, since this enlarges the reactive surface, thus promoting the development of acid-generation reactions [5].
The world consumption of mineral resources is very large and produces a large amount of waste, which has been reported as at least 5 to 7 billion tons of mine tailings per year [6], and up to 14 billion Sample OS-2 was selected for experiments because it represents most of the tailing material. This material was powdered and quartered to obtain homogeneous samples. According to the chemical composition of the raw tailing, two glasses were prepared, taking into account that the composition should be equilibrated with respect to the content of former, stabilizer, and refining oxides to produce a commercial quality glass [17]. To this end, the tailing sample was mixed with Na2CO3 (PANREAC, Barcelona, Spain cod.131648). Two glasses were obtained in order to establish their industrial behaviour by using 80 wt.% tailing and 20 wt.% Na2CO3 (sample G90-10), and 90 wt.% tailing and 10 wt.% Na2CO3 (sample G80-20). In order to obtain a homogeneous sample, each mixture was homogenized in a ball mill for 48 h.

Analytical Methods
The particle size distribution of the tailing samples was determined by means of an LS 13 320. Coulter particle size analyser (Beckman, Brea, CA, USA). The chemical composition of tailing samples was determined by X-ray fluorescence (XRF) using a Thermo Scientific ARL ADVANT'XP series instrument (ThermoFisher Scientific, Waltan, MA, USA). The mineralogical composition of the tailings and the phases formed during the thermal treatment of the glasses were determined by X-ray powder diffraction (XRD) using a Bragg-Brentano PANAnalyticalX'Pert Diffractometer. The system has graphite monochromator, an automatic gap, Cu Kα-radiation at λ = 1.54061 Å, powered at 45 kV-40 mA, a scanning range of 4-100° with a 0.017° 2θ step scan and a 50 s measuring time. Identification and semiquantitative evaluation of the phases was performed by using PANanalytical X'Pert HighScore software (Malvern PANanalytical, Malvern, UK). XRD and scanning electron microscopy (SEM) were used to demonstrate the amorphous structure of the glasses. The equipment used was a compact Phenom XL Desktop SEM (ThermoFisher Scientific, Waltan, MA, USA).
The density of the glass was measured by the Archimedes method using toluene as the immersion liquid following the Spanish standard UNE-EN 993-2:1996 [30].
Thermal analysis of samples was obtained by simultaneous differential thermal analysis and thermogravimetry (DTA-TG), using a Netzsch instrument (STA 409C model NETZSCH-Gerätebau Three composite samples of about 10 kg each were collected from profiles along different levels of height of the tailing. Samples OS-1 and OS-3 represent the lower and uppermost parts of the tailing and OS-2 was obtained at the intermediate height of the tailing. Samples were dried, sieved and quartered to determine their chemical and mineralogical compositions. Sample OS-2 was selected for experiments because it represents most of the tailing material. This material was powdered and quartered to obtain homogeneous samples. According to the chemical composition of the raw tailing, two glasses were prepared, taking into account that the composition should be equilibrated with respect to the content of former, stabilizer, and refining oxides to produce a commercial quality glass [17]. To this end, the tailing sample was mixed with Na 2 CO 3 (PANREAC, Barcelona, Spain cod.131648). Two glasses were obtained in order to establish their industrial behaviour by using 80 wt.% tailing and 20 wt.% Na 2 CO 3 (sample G 90-10 ), and 90 wt.% tailing and 10 wt.% Na 2 CO 3 (sample G 80-20 ). In order to obtain a homogeneous sample, each mixture was homogenized in a ball mill for 48 h.

Analytical Methods
The particle size distribution of the tailing samples was determined by means of an LS 13 320. Coulter particle size analyser (Beckman, Brea, CA, USA). The chemical composition of tailing samples was determined by X-ray fluorescence (XRF) using a Thermo Scientific ARL ADVANT'XP series instrument (ThermoFisher Scientific, Waltan, MA, USA). The mineralogical composition of the tailings and the phases formed during the thermal treatment of the glasses were determined by X-ray powder diffraction (XRD) using a Bragg-Brentano PANAnalyticalX'Pert Diffractometer. The system has graphite monochromator, an automatic gap, Cu Kα-radiation at λ = 1.54061 Å, powered at 45 kV-40 mA, a scanning range of 4-100 • with a 0.017 • 2θ step scan and a 50 s measuring time. Identification and semiquantitative evaluation of the phases was performed by using PANanalytical X'Pert HighScore software (Malvern PANanalytical, Malvern, UK). XRD and scanning electron microscopy (SEM) were used to demonstrate the amorphous structure of the glasses. The equipment used was a compact Phenom XL Desktop SEM (ThermoFisher Scientific, Waltan, MA, USA).
The density of the glass was measured by the Archimedes method using toluene as the immersion liquid following the Spanish standard UNE-EN 993-2:1996 [30].
Thermal analysis of samples was obtained by simultaneous differential thermal analysis and thermogravimetry (DTA-TG), using a Netzsch instrument (STA 409C model NETZSCH-Gerätebau GmbH, Selb, Bavaria, Germany). Analyses were carried out in the temperature range 25-1300 • C under an air atmosphere, at a constant flow rate of 80 mL/min, in a platinum crucible and at a heating rate of 10 • C/min. The amount of sample analysed was~75 mg. The aluminium oxide Perkin Elmer 0419-0197 was taken as reference.
Glass transition temperature analysis (T g ), the dilatometric softening point (T d ) and the expansion coefficient were measured with a Linseis horizontal dilatometer L76/1550 (Linseis Messgeräte GmbH, Selb, Bavaria, Germany) using a 2 cm annealing bar. The experiment was carried out from room temperature to 1300 • C at a heating rate of 10 • C/min. Viscosity at T g had a constant value, independent of composition, of 1012.3 Pa·s [31].
The experimental viscosity-temperature curves (η-T) were drawn using the fixed points defined by Scholze [32] for hot-stage microscopy (HSM) according to rule DIN 51730 [33] and T g (obtained with dilatometry). In this case, a quenching glass sample, with 3-mm-high test cylinders, was conformed with raw materials powdered under 45 µm and bound using a 1/20 solution of Elvacite ® in acetone, and conformed in a uniaxial press. Test cylinders were heated at a 5 • C/min rate from room temperature to 1500 • C in an air atmosphere. This process was recorded in pictures with ProgRes Capture Pro software (Jenoptik AG, Jena, Germany). Analysis was performed with Hot-Stage software, developed by the Departament de Llenguatges i Sistemes Informàtics, ETSEIB, UPC [34]. The output points were plotted in a graph and then fitted to the Vögel-Fulcher-Tammann (VFT) equation. The theoretical viscosity-temperature curves were obtained using the model defined by Fluegel [35].
Glass colour was measured with a Spectrophotometer CM-700d Konica-Minolta Investment Ltd. (Sensing Business Division, Shanghai, China) and calibrated according to international standards [36]. The spectrophotometer was fitted with a barium sulphate-coated integrating sphere, and a standard illuminant C was used as a light source. The description of the colour was based on three parameters: lightness, saturation and intensity [37], which were based on the use of three coordinates, L*, a* and b*. L* indicates lightness, 100 being white colour and 0 black. The variables a* and b* are the chromatic coordinates: +a* is the red axis, −a* is the green axis, +b* is the yellow axis and −b* is the blue axis. Low L* values indicate that the glasses are dark and the a* and b* parameters correspond to yellowish green glasses. The colour purity was measured by the Chroma (C) parameter, which is defined by the equation: The intensity of colour is defined by the hue angle (h), which can be determined by Equation (2). Measurement range is 360 • , counted clockwise from 0 • for the red colour (a*).
The chemical stability of the glasses was determined by leaching tests in an acidic solution according to the DIN38414-S4 standard [38]. Leachates were analysed using inductively coupled plasma optical emission spectrometry (ICP-OES, Optima 3100, PerkinElmer, Waltham, MA, USA) and ICP-mass spectrometry (ICP-MS, Elan 6000, PerkinElmer, Waltham, MA, USA).

Glass Production
The raw mixtures were introduced in an alumina-mullite crucible and heated using a globular alumina furnace equipped with molybdenum disilicide Super Kanthal ® and a Eurotherm ® 902 programmer. Heating started at 2 • C min −1 up to 800 • C, followed by a holding time of 2 h, and a second step of heating at 1 • C min −1 up to 900 • C, in order to control the Na 2 CO 3 decomposition, and a rate of 2 • C min −1 up to 1450 • C, where it remained for 4 h. After this, the components were molten. Part of this melt was cast into a metallic mould preheated at 350 • C and annealed near the glass transition temperature (T g ) at 450 • C for 30 min, followed by free cooling inside the kiln until room temperature was reached. Another part of the melt was quenched via casting on a Cu plate.

Characterization of the Tailing Materials
The chemical composition of tailing materials from Osor has small variations, as shown in the analyses of the three samples from different heights of the tailing ( Table 1). The differences were not important for preventing their use in a glass-making plant, but the composition needs to be periodically controlled to adjust the additives. The more significant variation was in the SiO 2 content, which varied from 61.76 to 66.11 wt.%. The high calcium content of the tailings, about 12 wt.%, rendered the addition of this component unnecessary. However, the low content in alkaline oxides, 1.33-1.54 wt.% of Na 2 O and 2.46-2.66 wt.% of K 2 O, rendered the addition of them necessary. The content of potentially toxic elements was significant; the experimental sample had As 12.6, Bi 0.212, Sb 3.1, Hg 0.064, Cr 35.5, Pb 4930, Zn 4500 and Cu 264 ppm.
The mineralogical composition of the Osor tailing showed small variations in different areas. The most abundant minerals were quartz, albite, K-feldspar, muscovite, chlorite and calcite ( Figure 2). Locally, gypsum and sphalerite occurred in relatively significant amounts (Table 2).

Characterization of Glasses
The XRD diagrams (Figure 3a) show no peaks, indicating the amorphous structure of the glasses produced. The absence of crystalline phases was also tested by SEM observations, which show uniform composition (Figure 3b,c).

Characterization of Glasses
The XRD diagrams (Figure 3a) show no peaks, indicating the amorphous structure of the glasses produced. The absence of crystalline phases was also tested by SEM observations, which show uniform composition (Figure 3b,c). In both the glasses produced (Table 3), only the Na2O content varied: 7.55 wt.% in the glass made with 90 wt.% of tailing and 14.1 wt.% Na2O in that made with 80% of tailing. When the Al/Na and 2Al/Ca ratios are ≤1, Al plays a role of former element and is located in a tetrahedral site [39]. In the glasses produced, Al/Na was 0.52 and 0.27, and 2Al/Ca was 0.69. In this case, at least part of the alkaline and alkaline earth elements acted as charge compensators [23]. This composition is suitable In both the glasses produced (Table 3), only the Na 2 O content varied: 7.55 wt.% in the glass made with 90 wt.% of tailing and 14.1 wt.% Na 2 O in that made with 80% of tailing. When the Al/Na and 2Al/Ca ratios are ≤1, Al plays a role of former element and is located in a tetrahedral site [39]. In the glasses produced, Al/Na was 0.52 and 0.27, and 2Al/Ca was 0.69. In this case, at least part of the alkaline and alkaline earth elements acted as charge compensators [23]. This composition is suitable for producing durable glass and also for obtaining glass-ceramics; thus, the SiO 2 + Al 2 O 3 content was between 60 and 70 wt.% [40]. On the other hand, the high fluorine content made it possible to reach a melting state at a moderate temperature. The density values of the produced glasses were 2.54 g/cm 3 for the G 80-20 glass and 2.61 g/cm 3 for the G 90-10 glass, which are slightly higher values than those of a soda-lime glass (2.44 g/cm 3 ).

Thermal Behaviour
The DTA of the two samples show glass-transition temperatures (T g ) of 670 and 660 • C for G 80-20 and G 90-10 , respectively (Figure 4). The exothermic peak corresponds to the formation of crystalline phases. This event occurred at a slightly lower temperature, 875 • C, and was more pronounced in the G 80-20 glass, whereas in the G 90-10 glass it occurred at 901 • C. Two endothermic peaks correspond to the melting temperature (T m ) of the two crystalline phases formed during the thermal treatment of the glasses. T m was differed considerably for the glasses: 1066 and 1096 • C in the G 90-10 glass and 1215 and 1220 • C in the G 80-20 glass. The density values of the produced glasses were 2.54 g/cm 3 for the G80-20 glass and 2.61 g/cm 3 for the G90-10 glass, which are slightly higher values than those of a soda-lime glass (2.44 g/cm 3 ).

Thermal Behaviour
The DTA of the two samples show glass-transition temperatures (Tg) of 670 and 660 °C for G80-20 and G90-10, respectively (Figure 4). The exothermic peak corresponds to the formation of crystalline phases. This event occurred at a slightly lower temperature, 875 °C, and was more pronounced in the G80-20 glass, whereas in the G90-10 glass it occurred at 901 °C. Two endothermic peaks correspond to the melting temperature (Tm) of the two crystalline phases formed during the thermal treatment of the glasses. Tm was differed considerably for the glasses: 1066 and 1096 °C in the G90-10 glass and 1215 and 1220 °C in the G80-20 glass. The Hrubý coefficient, Hr, [41] provides a useful criterion to evaluate the stability of glasses [42]. This coefficient is defined by Equation (3): where Tc is the crystallization temperature, Tg is the glass transition temperature determined by the DTA, Tc is the main temperature of crystallization and Tm is the beginning of the melting. Hr The Hrubý coefficient, Hr, [41] provides a useful criterion to evaluate the stability of glasses [42]. This coefficient is defined by Equation (3): where T c is the crystallization temperature, T g is the glass transition temperature determined by the DTA, T c is the main temperature of crystallization and T m is the beginning of the melting. Hr inversely correlates with the tendency to crystallize the glass. The Hr of the G 90-10 and G 80-20 glasses is 1.35 and 0.85, respectively, which is indicative of high stability with respect to crystallization, especially in the glass obtained with 90 wt.% of tailing. The stability of glasses is important when they contain potentially toxic elements that must be immobilized. Thus, poor stability may lead to the formation of crystals either during annealing of the glass, or afterwards, and lead to cracking of the glass, which increases the leaching rates in such glasses [43]. The higher soda content of the G 80-20 glass significantly decreases working temperatures (forming, conditioning and melting).
The crystalline phases formed during the crystallization of glasses at 1000 • C were different in both glasses, as suspected by the difference between their T c , which was determined from the DTA ( Figure 5).
The stability of glasses is important when they contain potentially toxic elements that must be immobilized. Thus, poor stability may lead to the formation of crystals either during annealing of the glass, or afterwards, and lead to cracking of the glass, which increases the leaching rates in such glasses [43]. The higher soda content of the G80-20 glass significantly decreases working temperatures (forming, conditioning and melting).
The crystalline phases formed during the crystallization of glasses at 1000 °C were different in both glasses, as suspected by the difference between their Tc, which was determined from the DTA ( Figure 5). The G90-10 glass crystallized at higher temperature and the phases formed were richer in Ca, being 88 wt.% of intermediate plagioclase and 12 wt.% of crystals from the akermanite-gehlenite (Ca2Al(AlSiO7)-Ca2Mg(Si2O7)) solid solution. In the G80-20 glass, crystalline phases were 66 wt.% nepheline, 25 wt.% plagioclase and 9 wt.% diopside. The type of plagioclase formed depends on the Ca/Na ratio; with a high content of Na, the crystallization of an intermediate or a Na-rich plagioclase is expected. Therefore, this occurred at a lower temperature than in the Ca-predominant composition, where plagioclase with a composition near the anorthite end member of the plagioclase solid solution (NaCaAlSi3O8-Ca2Al2Si2O8) crystallized.

Rheological Properties
The dilatometric analysis of the two glasses ( Figure 6) yielded a Tg of 637 °C in the G80-20 glass and 628 °C in the G90-10 glass. The comparison of these values with those obtained by DTA-TG shows a difference of about 30 °C between both.
Both rheological properties described above were slightly higher in the Osor glasses than in a typical soda-lime glasses. These commercial glasses have a Tg from 550 to 580 °C and an expansion coefficient between 8 × 10 -6 and 9.00 × 10 -6 °C −1 [45]. The G 90-10 glass crystallized at higher temperature and the phases formed were richer in Ca, being 88 wt.% of intermediate plagioclase and 12 wt.% of crystals from the akermanite-gehlenite (Ca 2 Al(AlSiO 7 )-Ca 2 Mg(Si 2 O 7 )) solid solution. In the G 80-20 glass, crystalline phases were 66 wt.% nepheline, 25 wt.% plagioclase and 9 wt.% diopside. The type of plagioclase formed depends on the Ca/Na ratio; with a high content of Na, the crystallization of an intermediate or a Na-rich plagioclase is expected. Therefore, this occurred at a lower temperature than in the Ca-predominant composition, where plagioclase with a composition near the anorthite end member of the plagioclase solid solution (NaCaAlSi 3 O 8 -Ca 2 Al 2 Si 2 O 8 ) crystallized.

Rheological Properties
The dilatometric analysis of the two glasses ( Figure 6 Viscosity is an important property of the raw glass melt for glass making. The control of this parameter is essential in commercial production for achieving a high efficiency and quality of products [46]. The experimental fixed viscosity points obtained by the automatic image analysis provided by HSM are given in Table 4. Figure 7 shows the images corresponding to the thermal geometrical evolution of the G80-20 glass.
According to the working range, glasses can be classified as short glasses and long glasses. The working range of the temperature has been established on viscosity values between 10 6 and 10 3 Pa·s [46] or between 10 4 and 10 8 Pa·s [47]. Long glasses present a working range of than 400 °C [47]. The forming and shaping of long glasses is easier and allows for working at lower temperatures, while the short glasses become rigid more quickly, enabling automatized production processes [47,48]. The viscosity range between 10 6 and 10 3 Pa·s in the case of G90-10 was 293 °C and in G80-20 was 479 °C. Therefore, the manufacture of glasses can only be carried out automatically in the case of raw tailing with a higher soda addition.  Calculated temperatures of the significant production viscosities b Lower annealing point/10 13.5 620 585 Figure 6. Thermal expansion plot as a function of temperature used to determine T g and T d .
Both rheological properties described above were slightly higher in the Osor glasses than in a typical soda-lime glasses. These commercial glasses have a T g from 550 to 580 • C and an expansion coefficient between 8 × 10 −6 and 9.00 × 10 −6 • C −1 [45].
Viscosity is an important property of the raw glass melt for glass making. The control of this parameter is essential in commercial production for achieving a high efficiency and quality of products [46]. The experimental fixed viscosity points obtained by the automatic image analysis provided by HSM are given in Table 4. Figure 7 shows the images corresponding to the thermal geometrical evolution of the G 80-20 glass. The viscosity-temperature curves for the glasses produced ( Figure 8) were obtained using HSM results and by determining the correlation between the fixed points of the known viscosity, the Tg obtained by a dilatometer corresponding to a viscosity of 10 12.3 Pa·s and the Vogel-Fulcher-Tammann (VFT) equation [32]. The G80-20 curve has a relatively low fitting in the intermediate region of viscosity, which can be attributed to the crystallization stage [50].

Optical Properties
The CIEL AB space [36] provides an objective measure of the colour. In Table 5, the colour parameters for the glasses from the Osor tailings are shown. The low values of C* indicate that the glasses plotted near the centre of the colour sphere and thus, present low purity. Hue values of the Osor glasses were 93.18 and 110.32°; values between 90 and 180° indicate yellow-green tonality. In According to the working range, glasses can be classified as short glasses and long glasses. The working range of the temperature has been established on viscosity values between 10 6 and 10 3 Pa·s [46] or between 10 4 and 10 8 Pa·s [47]. Long glasses present a working range of than 400 • C [47]. The forming and shaping of long glasses is easier and allows for working at lower temperatures, while the short glasses become rigid more quickly, enabling automatized production processes [47,48]. The viscosity range between 10 6 and 10 3 Pa·s in the case of G 90-10 was 293 • C and in G 80-20 was 479 • C. Therefore, the manufacture of glasses can only be carried out automatically in the case of raw tailing with a higher soda addition. 1318-1583 1282-1420 Melting range/10 2 -10 1 >1583 >1420 Workability interval/10 5 -10 2 1130-1583 1066-1420 a Engels and Link [31] included the glass transition point by dilatometer. b Calculated using the model by Fluegel [35].
The viscosity-temperature curves for the glasses produced ( Figure 8) were obtained using HSM results and by determining the correlation between the fixed points of the known viscosity, the T g obtained by a dilatometer corresponding to a viscosity of 10 12.3 Pa·s and the Vogel-Fulcher-Tammann (VFT) equation [32]. The G 80-20 curve has a relatively low fitting in the intermediate region of viscosity, which can be attributed to the crystallization stage [50]. The viscosity-temperature curves for the glasses produced ( Figure 8) were obtained using HSM results and by determining the correlation between the fixed points of the known viscosity, the Tg obtained by a dilatometer corresponding to a viscosity of 10 12.3 Pa·s and the Vogel-Fulcher-Tammann (VFT) equation [32]. The G80-20 curve has a relatively low fitting in the intermediate region of viscosity, which can be attributed to the crystallization stage [50].

Optical Properties
The CIEL AB space [36] provides an objective measure of the colour. In Table 5, the colour parameters for the glasses from the Osor tailings are shown. The low values of C* indicate that the glasses plotted near the centre of the colour sphere and thus, present low purity. Hue values of the Osor glasses were 93.18 and 110.32°; values between 90 and 180° indicate yellow-green tonality. In Figure 9, these parameters are shown together with others obtained from glasses produced using tailings from different origins, whose characteristics were previously published [13,19,51]. Comparison of the colour parameters from these glasses indicates that tailing glasses usually show medium or dark colour and a low purity. These results are due to the abundant metals content, or

Optical Properties
The CIEL AB space [36] provides an objective measure of the colour. In Table 5, the colour parameters for the glasses from the Osor tailings are shown. The low values of C* indicate that the glasses plotted near the centre of the colour sphere and thus, present low purity. Hue values of the Osor glasses were 93. 18 and 110.32 • ; values between 90 and 180 • indicate yellow-green tonality. In Figure 9, these parameters are shown together with others obtained from glasses produced using tailings from different origins, whose characteristics were previously published [13,19,51]. Comparison of the colour parameters from these glasses indicates that tailing glasses usually show medium or dark colour and a low purity. These results are due to the abundant metals content, or the chromophore elements, which usually occurs in tailings from metallic deposits. Yellow-green glasses are common when tailings contain high iron levels and low Mn/Fe ratios [52]. The refractive index, n, of both glasses produced was 1.54, which is similar than that of typical soda-lime-silica commercial glasses, with an n of 1.51 to 1.52, and slightly higher than in vitreous glasses because cations fill the interstices in the network [45].

Leaching Behaviour
The test for leachability of potentially toxic elements from the glasses of the Osor tailings, determined using the DIN 38414-S4 standard [36], showed that the concentrations of metals in the leachates were always under the threshold limits established in the standard [36]. These results suggest that the metals were fixed in the structure of the glasses. Metal retention was similar in both glasses, although the G80-10 glasses better retained metals. The Pb content in the G90-10 glass was 4437 ppm, from this, 0.230 ppm was removed during the leaching test, which represents 0.0005 wt.% of the total amount of Pb in this glass (Table 6). These results indicate the high stability of the toxic elements in the glassy structure, as has been reported in other glasses made from tailings [13,19,51] and in industrial waste products [53].  The refractive index, n, of both glasses produced was 1.54, which is similar than that of typical soda-lime-silica commercial glasses, with an n of 1.51 to 1.52, and slightly higher than in vitreous glasses because cations fill the interstices in the network [45].

Leaching Behaviour
The test for leachability of potentially toxic elements from the glasses of the Osor tailings, determined using the DIN 38414-S4 standard [36], showed that the concentrations of metals in the leachates were always under the threshold limits established in the standard [36]. These results suggest that the metals were fixed in the structure of the glasses. Metal retention was similar in both glasses, although the G 80-10 glasses better retained metals. The Pb content in the G 90-10 glass was 4437 ppm, from this, 0.230 ppm was removed during the leaching test, which represents 0.0005 wt.% of the total amount of Pb in this glass (Table 6). These results indicate the high stability of the toxic elements in the glassy structure, as has been reported in other glasses made from tailings [13,19,51] and in industrial waste products [53].

Conclusions
Tailings from the Osor mines are rich in calcium and only soda needs to be added to make commercial glass.
Two different doses of Na 2 CO 3 were used for the experimental glasses. The transition and melting temperatures were higher in the case glasses with less soda content. The Hr of G 90-10 was higher than that of G 80-20 , suggesting that glass with less soda content was more stable with respect to crystallization. However, G 80-20 displayed a wider range of workability temperatures. Therefore, glasses were more stable, but more difficult to produce in the compositions with 10% Na 2 CO 3 compared to those with 20% Na 2 O 3 .
The composition of the tailings from the Osor mines is suitable for the manufacture of commercial glass. The glasses obtained from the tailings retain potentially toxic elements in their structure and prevent environmental contamination. Although both compositions retain these potentially toxic elements, glass with 20% Na 2 CO 3 is more efficient.
The tailing and Na 2 O 3 dosages are suitable for making commercial glass. Although the higher Na 2 O 3 content resulted in greater efficiency, the cost of production should also be taken into account before a final decision is made.