It can be seen in Fig. 4 that the separation of slag and iron after carbon thermal reduction reaction is very good, and a relatively regular cake shaped iron block was obtained, the testing and calculation of iron also indicates that the recovery rate of iron in desulfurization slag is higher than 99%. XRD spectra of reducing slag in Fig. 5 show that the main crystalline phase is diopside and the secondary crystalline phase is nepheline. The reducing slag was heated to molten state and then homogenized to prepare basic glass, simulating the high-temperature melting slag directly preparing basic glass and glass-ceramics, the macroscopic photos are shown in Fig. 6. The glass-ceramic process of basic glass usually includes two processes, namely the formation of crystal nucleus (endothermic process) and the growth of crystal (exothermic process). Therefore, endothermic peaks and exothermic peaks appear in the differential temperatures in DSC curve. As can be seen from the DSC curve in Fig. 7, sample 1 has an obvious endothermic peak around 985 K, and two obvious exothermic peaks around 1040 K and 1155 K. In comparison, the endothermic peak are not obvious in sample 2, 3, and 4, in additional, the temperatures of former exothermic peak drops to about 1030 K and the intensity of the peaks also becomes remarkably smooth, even indistinguishable, while the latter exothermic peak also drops to around 1130 K and the intensity of the peaks was enhanced, the DSC curve indicated that two crystal phases might precipitate in each of the four samples, and the crystal phase near 1130 K is obviously more than that near 1030 K for sample 2, 3, and 4 .
With the increase of heating rate, both the initial crystallization temperature and the peak temperature of the four samples increased gradually, which was mainly caused by the heat cannot be supplied in time with the increase of heating rate[17]. From the position of endothermic and exothermal peak, these two temperatures are far lower for that the glass-ceramic’s nucleation and crystallization temperatures of metallurgical slag-based are usually about 900 and 1100 ℃[16], respectively. It can be seen that the formation of nepheline crystals is conducive to reducing the heat treatment temperature and reducing the process energy consumption.
Based on the DSC curves of the four samples, it was determined that the nucleation was carried out at 1033 K for 2 h, and the crystallization was carried out at 1138 K for 2 h, with a heating rate of 5 ℃/min, and the glass-ceramics obtained after the heat treatment are shown in Fig. 6. XRD analysis of the glass-ceramics in Fig. 5 shows that the crystalline phases are diopside and nepheline, which is consistent with the XRD spectra of the reduction slag and the results of the two exothermic peaks in the DSC curve. However, from crystal phase intensity of diopside and nepheline, the difference between the four samples was small, which seemed to be not consistent with the characteristics of DSC curves. In fact, 1033 K was not only the formation point of crystal nucleus, but also the growth point of nepheline near 1030 K. In other words, after holding for 2 h at 1033 K in heat treatment, it also promotes the crystallization of nepheline.
The glass-ceramics is obtained by the controlled crystallization of the basic glass after heat treatment, which is a non-uniform nucleation process. At present, classical theories on the crystallization behavior of glass-ceramics mainly include Kissinger equation (Eq. 3), Ozawa equation (Eq. 4) and Augis-Bennett equation (Eq. 5). In this article, these three formulas were used to calculate the crystallization kinetics of glass-ceramics under non-isothermal conditions. The crystallization activation energy E was used to represent the potential barrier needed to overcome the structural rearrangement during the transition from glassy state to crystal state, that is, the difficult degree of crystallization, and the crystal growth index n was used to judge the growth form of crystal.
ln(Tp2⁄α) = E⁄(R×Tp) + ln(E⁄R)-lnv (3)
ln(1⁄α) = E⁄(R×Tp) + C (4)
ln(Tp⁄α) = E⁄(R×Tp) + lnK0) (5)
n=(2.5×R×Tp2)⁄(∆T×E) (6)
where α is heating rate, Tp is the peak temperature of crystallization, v is the frequency factor, C is constant, ∆T is the half peak width of the crystallization peak.
According to the DSC curves of the four samples at different heating rates, combined with the Kissinger equation, Ozawa equation and Augis-Bennett equation, the linear fitting was carried out with 1/Tp as the X-axis and ln(Tp2/α), ln(1/α) and ln(Tp/α) as the Y-axis, respectively. The linear fitting of dynamics was obtained in Fig. 8. Then Tp, α and R are substituted into the equation to obtain the crystallization activation energy E, as shown in Table 2. By comparing and analyzing the crystallization activation energy obtained by different methods, it can be found that there is little difference between results obtained by the three methods. The value of Ozawa method is relatively high, while the value of Kissinger method is relatively low. Compared with the four samples, the crystallization activation energy of sample No. 3 is the smallest, that is, sample No. 3 has the lowest potential barrier needed to overcome the structural rearrangement and is easier to crystallization.
Table 2
Crystallization activation energy of different samples
Sample
|
Tp(K)
|
Crystallization activation energy E
|
5
|
10
|
15
|
20
|
Kissinger
|
Ozawa
|
Augis-Bennett
|
Average
|
1
|
1143.75
|
1152.75
|
1160.75
|
1164.85
|
692.5
|
711.7
|
702.2
|
702.1
|
2
|
1111.15
|
1124.05
|
1130.25
|
1137.65
|
542.3
|
561.0
|
551.7
|
551.7
|
3
|
1110.55
|
1128.35
|
1137.85
|
1148.65
|
374.4
|
393.2
|
383.8
|
383.8
|
4
|
1125.65
|
1137.15
|
1143.05
|
1149.95
|
607.3
|
626.2
|
616.8
|
616.8
|
The crystal growth index was calculated by using the Augis-Bennett equation (Eq. 6). By substituting the E values of different crystallization activation energies calculated in Table 2 into the equation, and combining with the DSC curve, the crystal growth index n shown in Table 3 can be obtained. It can be seen that the crystal growth index is not very high, which may be due to the fact that there are few oxides with nucleation function, such as TiO2, in the raw material under the condition of not adding nucleating agent in this paper, and two kinds of crystals are precipitated during heat treatment, which results in mutual interference in the initial nucleation stage and weakened crystallization ability of glass. Among them, the n value of sample No. 3 is the largest at the heating rate of 5 K/min. The crystallization mode of the glass-ceramics is changed from two-dimensional to one-dimensional volume crystallization[17], indicating that the preparation of glass-ceramics in this paper needs to add nucleating agent.
Table 3
Crystal growth index of different samples
Sample
|
Heating rate
|
Half peak width ∆T
|
Kissinger
|
Ozawa
|
Augis-Bennett
|
Average
|
1
|
5
|
24.1
|
1.6
|
1.6
|
1.6
|
1.6
|
10
|
24.0
|
1.7
|
1.6
|
1.6
|
1.6
|
15
|
23.0
|
1.8
|
1.7
|
1.7
|
1.7
|
20
|
28.1
|
1.4
|
1.4
|
1.4
|
1.4
|
2
|
5
|
28.1
|
1.7
|
1.6
|
1.7
|
1.7
|
10
|
47.1
|
1.0
|
1.0
|
1.0
|
1.0
|
15
|
46.1
|
1.1
|
1.0
|
1.0
|
1.0
|
20
|
52.8
|
0.9
|
0.9
|
0.9
|
0.9
|
3
|
5
|
24.7
|
2.8
|
2.6
|
2.7
|
2.7
|
10
|
34.6
|
2.0
|
1.9
|
2.0
|
2.0
|
15
|
34.8
|
2.1
|
2.0
|
2.0
|
2.0
|
20
|
44.4
|
1.6
|
1.6
|
1.6
|
1.6
|
4
|
5
|
40.6
|
1.1
|
1.0
|
1.1
|
1.1
|
10
|
41.7
|
1.1
|
1.0
|
1.0
|
1.0
|
15
|
42.5
|
1.1
|
1.0
|
1.0
|
1.0
|
20
|
46.9
|
1.0
|
0.9
|
1.0
|
1.0
|