Zhitao Baia,
Guibo Qiub,
Ben Pengb,
Min Guoa and
Mei Zhang*a
aSchool of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, P. R. China. E-mail: zhangmei@ustb.edu.cn; Fax: +86 1062334926; Tel: +86 1062334926
bCentral Research Institute of Building and Construction Co., Ltd, MCC Group, Beijing, China
First published on 24th May 2016
Glass-ceramics have been successfully prepared from high-carbon ferrochromium slag (HCFS) and waste glass (WG), and the microstructural characterization and mechanical properties of the glass-ceramics were subsequently investigated. The development of HCFS-based glass-ceramics involves the nucleation and crystallization stages from the parent glass. With the increase in mass ratio of HCFS and WG (R(H/W)) from 0.60 to 1.67, the number of bridging oxygens of Si in the parent glass is reduced, as shown via Raman spectroscopy. Thus, their degree of polymerization decreases with it, and the temperature of nucleation and crystallization increase, which is consistent with the DSC results. The SEM images and EDS results indicate that the increasing value of R(H/W) decreases the crystal grain size and consequently increases the microhardness of the glass-ceramics. But the porosity simultaneously increases, which makes the bending strength increase at first and subsequently decrease. And the optimum properties of HCFS-based glass-ceramic samples in the present work are obtained when R(H/W) reaches 1.29, that is, a bending strength of 104 MPa and a microhardness of 9860 MPa.
Glass-ceramics are fine-grained polycrystalline materials that are crystalized after heat-treating the glass phase of suitable compositions. Commonly, glass-ceramics are not fully crystalline, that is, they are approximately 50–95 vol% crystalline phase with residual glass. Glass-ceramics have superior mechanical properties compared to the parent glass. At present, glass-ceramic materials have been successfully prepared from solid waste. The commercial exploitation of solid waste, particularly blast furnace slag,15–19 is well developed because the chemical compositions of blast furnace slag are stable and can be easily made into glass-ceramics.20–22 The HCFS and blast furnace slag have similar compositions, such as Al2O3, SiO2, CaO and MgO, which consist more than 90% content of the slag, so that the HCFS is reasonably suggested to synthesize glass-ceramics. A review of the phase analysis reveals that HCFS contains too many high-melting phases and requires high temperatures to melt, and therefore, some composition adjustments and sintering aids need to be involved, and complex reactions must be considered. According to the present study, there is limited research about synthesizing glass-ceramics from HCFS. The work of Wang et al.23,24 preliminarily studied the preparation of glass-ceramics from ferrochrome slag and indicated that when the percentage of slag reached 50 wt% in the raw materials, the slag-based glass-ceramic had a maximum microhardness of 8000 MPa. Thus, this study is motivated to comprehensively use HCFS. The characterizations of HCFS are investigated, and a HCFS-based glass-ceramic is synthesized.
To produce glass-ceramics at an appropriate temperature that contains the high melting point of HCFS, the prime problem was to decrease the melting point of the glass-ceramics system. Hence, waste glass (WG) was selected for the composition adjustment, limestone and soda (≤75 μm) were used as the fining agent, and fluorite (≤75 μm) was added as the flux. All of the chemical compositions are shown in Table 1. Five samples with different mass ratios of HCFS and WG (R(H/W)) were investigated, and their compositions are shown in Table 2. The phase diagram of the glass-ceramics was calculated using Factsage 6.2 before its synthesis, and the results are shown in Fig. 2. The target phases were consisted of diopside, nepheline, and a small amount of pyroxene. The calculated melting point of this system is approximately 1500 °C. Hence, it is reasonable to suggest that the melting point of the glass-ceramics system can be decreased by composition adjustment from the thermodynamic calculation viewpoint.
Raw materials | SiO2 | MgO | Al2O3 | CaO | Na2O | Cr2O3 | Fe2O3 | K2O | CaF2 |
---|---|---|---|---|---|---|---|---|---|
HCFS | 29.61 | 37.88 | 22.79 | 2.06 | 0.14 | 4.89 | 1.69 | 0.11 | |
WG | 75.68 | 3.40 | 0.94 | 8.95 | 9.73 | 0.024 | 0.27 | 0.68 | |
CaCO3 | 55.44 | ||||||||
Na2CO3 | 58.37 | ||||||||
CaF2 | 98.5 |
Sample no. | HCFS | Waste glass | CaCO3 | Na2CO3 | CaF2 (extra addition) | Mass ratio of HCFS and WG, R(H/W) |
---|---|---|---|---|---|---|
1# | 30.0 | 50.0 | 15.0 | 5.0 | 5.26 | 0.60 |
2# | 35.0 | 45.0 | 15.0 | 5.0 | 5.26 | 0.78 |
3# | 40.0 | 40.0 | 15.0 | 5.0 | 5.26 | 1.00 |
4# | 45.0 | 35.0 | 15.0 | 5.0 | 5.26 | 1.29 |
5# | 50.0 | 30.0 | 15.0 | 5.0 | 5.26 | 1.67 |
![]() | ||
Fig. 2 Phase diagram of the CaO–MgO–SiO2–Na2O–Al2O3 system glass-ceramics, in which the Na2O content is 7.83–5.91%, and the Al2O3 content is 7.31–11.68%. |
Appropriate amounts of HCFS and WG were first ground to 75 μm and subsequently evenly mixed with the other materials, that is, limestone (CaCO3), soda (NaCO3) and fluorite (CaF2). The mixture was placed into a corundum crucible, melted at 1550 °C in a muffle furnace and kept for 2 hours. Then, the samples were poured into a stainless-steel mould that was pre-heated to 500 °C and annealed at 500 °C for one hour in a muffle furnace to produce the parent glass.
The parent glass was subsequently investigated with TG-DSC to determine the nucleation and crystallization temperatures. Then, a two-stage heat treatment was used to prepare the glass-ceramics, which included the nucleation and crystallization processes. First, the parent glass was heated from room temperature to the nucleation temperature with a step of 5 °C per minute and maintained for 2 hours. Afterwards, the samples were continuously heated to the crystallization temperature with steps of 5 °C per minute and subsequently maintained for 2 hours. Finally, the glass-ceramics were obtained by furnace cooling. The nucleation time and crystallization time are both set on the preliminary work. The parent glass have been treated from 30 to 150 min, the highest mechanical properties were achieved after 120 min nucleation and crystallization, and thus the nucleation time and crystallization time are both set at 2 hours in this work, which are also corresponded to the report.23,24
The mechanical properties of the HCFS-based glass-ceramics were determined using the microhardness and 3-point bending tests. Vickers (or Knoop) microhardness tests were conducted using a Leica-VMHT30M tester. Optical mounts were performed using standard metallographic techniques, and a load of 100–300 g was used to indent their surfaces. To obtain reliable statistical data, at least 13 indentations were made on each sample. The empirical relation of Ponton and Rawlings, which uses fracture lines that emanate from Vickers diamond indentation, was used to determine the fracture toughness values.25 Three-point bending tests were performed using a CDW-5, 5 KN tester with the sample size of 36 mm × 3 mm × 4 mm.
![]() | ||
Fig. 3 X-ray diffraction patterns of the HCFS-based parent-glass samples with different R(H/W) values. |
To analyse the molecular structures of the parent glass with different R(H/W), the samples were characterized by Raman spectra, and the results are shown in Fig. 4. The structural units in the silicate network are commonly described by the nomenclatures of Qi(Si), where i represents the number of bridging oxygen atoms per Si atom. It can be seen from Fig. 4(a) that all parent glass have three regions: 200–600 cm−1, 600–800 cm−1, and 800–1200 cm−1.
The peak at approximately 350 cm−1, which is associated with the bending motion of Si–O–Si,26,27 did not change with the increase in R(H/W), which reveals that the increasing content of HCFS does not affect the bending motion of Si–O–Si. The 600–800 cm−1 band, which is attributed to the presence of Si–O stretching linkages,28–31 became pronounced with the increase in R(H/W) because of the tiny increase in content of ringwoodite ((Mg, Fe)2SiO4) in the parent glass, as shown in Fig. 3. Meanwhile, the centre of this envelope curve shifted to a higher spectral frequency, which generally indicates a lower degree of polymerization of silicate structures. To further investigate the silicate structure, the Raman curves were deconvolved in the range of 800 to 1200 cm−1 according to the rules proposed by Mysen et al.27 It is assumed that the Raman curves are abided by Gaussian functions and that the bands are only fitted in the regions where obvious shoulders or peaks were observed or strictly proven by previous studies. Many studies32–34 have investigated the Raman spectra of silicate glasses and melts and assigned the peaks at approximately 870 cm−1, 960 cm−1, 990 cm−1, and 1050 cm−1 to the stretching vibrations of Q0(Si), Q1(Si), Q2(Si), and Q3(Si), respectively, each of which corresponds to a Gaussian fitting function. Then, four Gaussian functions were used in this study to fit the Raman curves of the parent-glass samples, and the fitting results are shown in Fig. 4(b)–(f).
According to the studies of Mysen and Frantz,35,36 the calculation of the mole fractions of different structural units, i.e., Qi(Si) (i = 0, 1, 2, and 3), could be calibrated by the Raman scattering coefficient, which is only related to Qi(Si) species. Therefore, the following expression can be used to calculate the mole fractions:
Xi = θiAi, i = 0, 1, 2, and 3 | (1) |
![]() | (2) |
The variation of the mole fractions of different Qi(Si) species with different R(H/W) values is shown in Fig. 5. The stretching vibrations of Q2(Si) and Q3(Si) decrease with the increase in R(H/W), but the stretching vibrations of Q1(Si) becomes higher, so that the number of bridging oxygen atoms of Si reduces. The number of bridging oxygen atoms of Si is the main factor that affects the degree of polymerization of the silicate structures in the melting glass-ceramics: fewer bridging oxygen atoms of Si in the system corresponds to a lower degree of polymerization. Thus, when R(H/W) increases, the degree of polymerization of the parent glass decreases, which may affect the nucleation and crystallization process and, particularly, their temperatures.
To determine the nucleation (Tn) and crystallization (Tg) temperatures, the partially enlarged DSC curves are shown in Fig. 6(b) and (c). As shown in Fig. 6(b), a series of small endothermic peaks that start from 630 °C are observed in each curve. The marks on each line are the starting temperature for the nucleation of glass-ceramics and defined as the Tn temperature to consider the lowest energy cost. There is a sharp exothermal peak in each DSC curve shown in Fig. 6(c), and the temperature of the peak is the best efficiency point of the crystallization, so it is defined as the Tg temperature. Thus, the optimum nucleation and crystallization temperatures of each sample are gathered and shown in Table 3. The nucleation and crystallization temperatures increase with the increase in R(H/W), which is attributed to the molecular structure of the parent glass. When R(H/W) increases, the number of bridging oxygen atoms of Si in the glass phase of the parent glass decreases according to the Raman spectra, as shown in Fig. 5. Thus, the heat and mass transfers become difficult, that is, the parent glass will require much more energy for nucleation and crystallization. Hence, the nucleation and crystallization temperatures must increase.
R(H/W) | 0.60 | 0.78 | 1.00 | 1.29 | 1.67 |
Nucleation temperature/°C | 635 | 642 | 648 | 652 | 658 |
Crystallization temperature/°C | 835 | 838 | 842 | 855 | 859 |
Considering the small-scale variation of the nucleation and crystallization temperatures and the large scale of content changing of HCFS in the raw materials of 30–50%, the occurrence of high-melting-point phases, such as ringwoodite and spinel, and the heat-treatment condition of the glass-ceramics in this study are obviously beneficial to large-scale industrial production.
Fig. 8 presents the backscattered electron images of the glass-ceramic samples with different R(H/W) values, and the EDS results of some randomly selected points are shown in Table 4. In Fig. 8, there are two phases in the glass-ceramics: white and grey phases. According the EDS results in Table 4, the white phase corresponds to CaMg(SiO3)2 and the grey phase is (Na, K)AlSiO4. A comparison of the XRD and SEM results shows that CaMg(SiO3)2 and (Na, K)AlSiO4 are the main phases, whereas Ca(Mg, Fe)Si2O6 and Ca(Mg, Al, Fe) (Al, Si)2O6 are the minor parts of the glass-ceramics. The grain size was counted from Fig. 8, and the results are shown in Fig. 9. With the increase in R(H/W), that is, the increasing content of HCFS in raw materials, the grain sizes of the white and grey phases decrease. This result is attributed to the increase in content of nucleation agent with the increase in the content of HCFS. Thus, the crystallization of the parent glass will become easier and quicker, which subsequently causes smaller crystal grain size.
![]() | ||
Fig. 8 SEM images of glass-ceramic samples with different R(H/W) values: (a) R(H/W) = 0.60; (b) R(H/W) = 0.78; (c) R(H/W) = 1.00; (d) R(H/W) = 1.29. |
Point no. | Element norm. C (wt%) | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
O | Mg | Al | Ca | Na | Cr | Fe | Si | Phase | ||
1 | Grey phase | 39.76 | 4.04 | 14.93 | 3.41 | 19.07 | 18.97 | (Na, K)AlSiO4 | ||
2 | 38.31 | 4.48 | 16.35 | 2.08 | 19.54 | 19.24 | ||||
3 | 39.42 | 4.76 | 14.74 | 4.38 | 18.14 | 18.57 | ||||
4 | White phase | 37.53 | 11.54 | 4.82 | 20.90 | 0.90 | 0.80 | 1.26 | 22.26 | CaMg(SiO3)2 |
5 | 37.40 | 11.71 | 4.63 | 20.07 | 2.14 | 0.75 | 1.14 | 22.16 | ||
6 | 36.53 | 11.48 | 4.46 | 21.70 | 1.29 | 0.65 | 1.27 | 22.62 |
Similarly, when the nucleating agent increases with the increase in the content of HCFS, the average crystal size decreases and crystallization accelerates. This phenomenon makes some gases unable to completely escape and become trapped inside the sample, which leads to a high degree of porosity, as shown in Fig. 10. In Fig. 11, the porosity of the glass-ceramics initially increases with the increase in R(H/W), but when R(H/W) is above 1.29, the porosity of the glass-ceramics suddenly increases.
![]() | ||
Fig. 10 Porosity measurement of glass-ceramic samples with different R(H/W) values: (a) R(H/W) = 0.78; (b) R(H/W) = 1.00; (c) R(H/W) = 1.29; (d) R(H/W) = 1.78. |
CaCO3 = CaO + CO2 | (3) |
Na2CO3 = Na2O + CO2 | (4) |
4CrFe4 + 11O2 = 16FeO + 2Cr2O3 | (5) |
2MgO + 2Cr2O3 = 2MgCr2O4 | (6) |
Chromium sesquioxide would further react with MgO and form the chrome spinel phase (eqn (6)). Then, the trace of high melting-point materials, that is, (Mg, Fe)2SiO4, MgAl2O4, Mg(Al, Cr)2O4, and Cr2O3, remains in the parent glass during the melting process and acts as nucleation sites. During the crystallization stage of the heat treatment, Fig. 7 and Table 4 show that CaO, MgO, FeO and SiO2 further react and form CaMg(SiO3)2 and Ca(Mg, Fe)Si2O6, as shown in eqn (7) and (8). The reactions among Na2O, K2O, Al2O3 and SiO2 are described in eqn (9) and form (Na, K)AlSiO4. These substances constitute the main phases of the glass-ceramics. (Na, K)AlSiO4 may also react with sylvite or sodium salt and generate a low-melting solid solution, which is likely the main source of the glassy phase.
CaO + MgO + 2SiO2 = CaO·MgO·2SiO2 | (7) |
CaO + MgO(FeO) + 2SiO2 = Ca(Mg, Fe)Si2O6 | (8) |
Na2O(K2O) + Al2O3 + 2SiO2 = 2(Na, K)AlSiO4 | (9) |
Fig. 12 is the representation of Vickers microhardness values of the HCFS-based glass-ceramic samples. The hardness values clearly continuously increase with the increase in R(H/W), possibly because of the increasing concentration of the nucleation agent. When R(H/W) increases, the average crystal size correspondingly decreases, as shown in Fig. 9, and further leads to the increase of the amount of grain boundary, which enhances the resistance to indentation. When the value of R(H/W) increases to 1.67, the microhardness of the glass-ceramic sample reaches a maximum of 9860 MPa, which is higher than that of other glass-ceramics made from metallurgical slags with a maximum value of 7000 MPa.25
Fig. 13 shows an error-bar representation of the bending strength of the HCFS-based glass-ceramic samples. Here, with the increase in R(H/W), the bending strength of the samples initially increases to the top (approximately 110 MPa) and subsequently slightly decreases. This phenomenon is attributed to the increase in the porosity of the specimens when the HCFS content increases, as shown in Fig. 10. When R(H/W) is beyond 1.29, the bending strength starts to decrease, whereas the porosity starts to sharply increase. Thus, a moderate porosity, that is, when the porosity is below 1.5%, has a notably small effect on the bending strength. However, an abundant porosity will deteriorate it. When the value of R(H/W) reaches 1.29, the bending strength of glass-ceramic sample reaches the maximum of 104 MPa. The bending strength of glass-ceramics that are made from metallurgical slag is generally 56.6–97 MPa,25,38−41 but some kind of the slag-based glass-ceramics that just focused on the enhancement of the bending strength has reached 136 MPa.25,42 Hence, the HCFS-based glass-ceramics in this study show a comparable bending strength with the normal slag-based glass-ceramics. In addition, HCFS-based glass-ceramic materials are produced from the toxic substances and wastes from the normal procedure. Thus, it is a good reference to recycle this kind of material because of its low-cost raw material, good properties and notably effective manner in treating toxic substances.
This journal is © The Royal Society of Chemistry 2016 |