1 Introduction

In China, Panzhihua, Xichang, and Chengde are commonly known for their rich resource of vanadium–titanium magnetite, which contains iron, vanadium, titanium, and other associated metals. However, traditional methods, such as “blast furnace and converter” involving titanomagnetite concentrate, are applicable to the recovery of iron and vanadium, but titanium is not considered in blast furnace slag. Furthermore, the application of blast furnace slag also failed to recover TiO2. So the TiO2 in vanadium–titanium magnetite is still lost in blast furnace slag and accounts for 600,000 tons per year [1]. Therefore, researches were prompted to develop new technologies that can be applied to separate and utilize iron, vanadium, and titanium cleanly and efficiently. Third-generation smelting technology, namely “direct reduction and electric furnace smelting,” was considerably developed. In this process, titanomagnetite concentrate is directly reduced with coal and then smelted in an electric furnace. Thus, the slag containing TiO2 approximately 40 %–50 % and direct reduction iron can be achieved. The TiO2 content of slag is twice as high as that of blast furnace slag. Moreover, the recovery rate of iron increases by 2 %–3 %; by comparison, the recovery rate of vanadium increases by more than 15 % with this technology. As a result, the TiO2 in direct reduction and smelting titanium slag can be comprehensively utilized [2].

Studies were focused on blast furnace slag [39]; however, limited information is available regarding the smelting titanium slag. To develop new process, researchers should consider the recovery of titanium in direct reduction and smelting titanium slag. Yang et al. [10] and Li [11] investigated the acidolysis of smelting titanium slag. Zhang and Zhong et al. [12, 13] performed hydrochloride leaching and “phosphate activation roasting and dilute sulfuric acid leaching” to produce acid-dissolved titanium slag. New utilization methods were also proposed to smelt titanium slag [1417]. Smelting titanium slag exhibits very complex properties; therefore, several drawbacks result from the direct application of this substance as a raw material in sulfuric acid method to produce titanium dioxide. For example, acidolysis conditions need high requirements, and concentrated acid wastes cannot be used. Impurities, including calcium, magnesium, aluminum, and silicon, are high and very difficult to eliminate; as such, coagulant dosage should be increased and time should be prolonged during acidolysis. Thus, production costs increase and productivity decreases. Nonferrous metals, including vanadium, chromium, and manganese, are also abundantly detected and difficult to eliminate. These elements affect the quality of titanium dioxide [18].

As a consequence, based on “selective enrichment, growth, and separation” theory proposed by Northeastern University [19], “titanium component growth-mineral separation” method was used to the smelting titanium slag. Titanium-bearing phase anosovite and gangue minerals contain fine crystal sizes; as such, these substances are also difficult to separate directly and effectively by beneficiation. Therefore, researchers should develop a reasonable and effective heat treatment process to promote growth and enhance the crystal properties of anosovite and gangue minerals. Mineral crystals should also be transformed from small to large sizes to obtain the necessary conditions for mineral separation. In this research, the processing mineralogy of smelting titanium slag and the effect of cooling rate on mineral phase reconstruction were investigated. This research aimed to recover titanium from smelting titanium slag.

2 Experimental

2.1 Materials

The materials used in the experiment were smelting titanium slag, which were provided by Panzhihua Iron and Steel Group Limited Company in China. They were cooled to room temperature directly from the molten state after electric furnace smelting. The processing mineralogy of smelting titanium slag was examined by chemical analysis, X-ray diffraction (XRD), and optical microscope (OM). Chemical analysis results, XRD patterns, and OM results are shown in Table 1 and Figs. 1 and 2, respectively.

Table 1 Chemical compositions of materials (wt%)
Fig. 1
figure 1

XRD pattern of materials

Fig. 2
figure 2

OM images of materials: a spinel (S) and anosovite (A) and b melilite (M)

Table 1 shows that the sample contains 46.8 % TiO2. The contents of iron, manganese, and sulfur are low. By contrast, the contents of harmful elements (vanadium and chromium) and impure elements (aluminum, silicon, calcium, and magnesium) are high. Vanadium and chromium affect the quality of titanium dioxide.

In Fig. 1, anosovite solid solution is the primary useful mineral and can be expressed as follows: m(AO·TiO2n(B2O3·TiO2), where A is the divalent titanium, magnesium, iron, manganese, etc., B is the trivalent titanium, iron, aluminum, chromium, etc., m and n are constant coefficient. Moreover, the raw materials contain spinel, hematite, and silicate minerals; among these minerals, anosovite exhibits the highest content, followed by spinel and silicate minerals. XRD patterns show that anosovite crystal is doped with magnesium.

OM images in Fig. 2 show that anosovite, spinel, and melilite coexist in the smelting titanium slag except the glass phase. Anosovite is brown and black and does not exhibit metallic luster. Anosovite mainly appears as an opaque material, but a small portion of this mineral is translucent. Anosovite is observed as columnar, tabular, acicular, and granular. Its content is approximately 50 %, and the particle size ranges from 0.02 to 0.20 mm.

2.2 Methods

The experiments were performed in a box-type resistance furnace. The heating element was molybdenum disilicide U-shaped bar. Temperature was controlled by an AI intelligent digital temperature controller with an accuracy of 2 K.

The smelting titanium slag was crushed to 2 mm using a crusher. Approximately 80 g of the sample was transferred into a graphite crucible and placed in a resistance furnace. The temperature of the furnace was increased to 1723 K and then maintained for 30 min. Then the resistance furnace was cooled to 1423 K at the cooling rates of 4.00, 2.00, 1.00, 0.50, and 0.25 K·min−1. Then the resistance furnace was closed, and the smelting titanium slag was cooled to room temperature slowly.

After the samples were subjected to heat treatment, the crystal volume and size of the minerals in smelting titanium slag were determined by OM. Mineral crystallization was analyzed using a D/Max2200 XRD analyzer (Rigaku, Japan). Surface micromorphological characteristics were examined using a Quanta 200 scanning electron microscopy (SEM, FEI, the Netherlands). Element contents were determined by Genesis 2000 EDS (EDAX, USA).

3 Results and discussion

3.1 Heating curve

Figure 3 illustrates the curve of temperature versus time during the melting process of the smelting titanium slag. After 280 min of heating, the smelting titanium slag reaches the temperature of 1723 K.

Fig. 3
figure 3

Heating curve of smelting titanium slag

3.2 OM analysis

After slow-cooling treatment, the samples were prepared as slices and glass slides. Then they were observed by OM to determine the volume and size of anosovite crystal. Semiquantitative analysis results are shown in Table 2. The anosovite crystal volume and size increase as cooling rate decreases. At the cooling rate of <2 K·min−1, crystal volume of 60 % and crystal size of 0.02–1.00 mm are observed, which indicate that slow-cooling treatment is beneficial to the increase in anosovite crystal volume and size.

Table 2 Semiquantitative analysis results of OM measurements

3.3 XRD analysis

Figure 4 shows the XRD patterns of the smelting titanium slag after the samples were subjected to different cooling rates. Cooling rate significantly affects mineral crystallization. Anosovite and spinel grow and become large in the slow-cooling process. Anosovite and spinel are developed completely, and the main characteristic peaks are evident at a cooling rate of 2 K·min−1. Moreover, the chemical formula of anosovite changes from Mg0.6Ti2.4O5 to MgTi2O5 (Table 3). Mg transfers to anosovite crystal, thereby increasing the stability of crystal lattice and promoting crystal growth.

Fig. 4
figure 4

XRD patterns of smelting titanium slag after different cooling rate treatments

Table 3 XRD results

3.4 SEM–EDS analysis

Figure 5 shows the SEM images and EDS spectra of the raw materials and the smelting titanium slag at 2 K·min−1. As can be seen from the SEM images, the minerals are observed mainly as granular with a sharp, and a small number take the form of subhedral tabular and acicular. EDS spectra show that titanium, magnesium, aluminum, silicon, calcium, and other elements in the raw materials are distributed. However, these elements are concentrated at 2 K·min−1, which means that the elements in the smelting titanium slag redistribute during the slow-cooling process. Anosovite and spinel crystals grow more completely and contain lower impure elements.

Fig. 5
figure 5

SEM images and EDS spectra: ad raw materials, eh smelting titanium slag at 2 K·min−1

3.5 Mechanism

In thermodynamics, crystallization begins when the melt cools to a temperature less than their corresponding melting points; at such temperatures, the free energy of the solid phase is lower than that of the liquid phase. As a result, the short-range regular structures of liquids form crystal embryos. Crystal embryos with different sizes, which exhibit the same structure as the solid phase, form in the melt. These crystal embryos also attract solute atoms to grow, form crystal nucleus with a certain size, and develop into full crystals. The whole crystallization process involves the formation and growth of crystal nucleus. Cooling rate significantly affects the shape, size, and purity of mineral crystals [20, 21].

3.5.1 Effect of cooling rate on crystal shape

Rapid cooling crystals often develop into columnar, acicular, and scaly structures; such crystals even grow into skeleton-like structures. Under unbalanced conditions, a high energy on crystal surfaces causes an unstable state as a crystal grows, resulting in the growth of skeleton-like crystals along edges or corners. If a crystal grows slowly under almost balanced conditions, a complete crystallized polyhedron usually forms.

3.5.2 Effect of cooling rate on crystal size

As crystallization occurs in the melt, a fast cooling rate causes the crystal to grow quickly; as a result, numerous crystal nuclei form. However, crystallization completes in a short time and numerous small crystals contain irregular granular aggregations. In addition, solute atoms have insufficient time to adhere to the crystal nucleus and promote crystal growth. Thus, aphanitic minerals and amorphous glass form.

During crystal growth, crystal nuclei absorb one another. Small crystals are also attracted by large ones if the crystal growth rate is slow. Therefore, only a small portion of the crystal nuclei continues to develop. In this case, crystals with ideal structures occupy a large volume.

3.5.3 Effect of cooling rate on crystal purity

Some materials in the melt are included in a crystal when crystals grow rapidly, thereby forming inclusions. Structural defects, which are caused by ion doping in the lattice, deteriorate the purity and integrity of crystals. Ideal crystals can be obtained when crystals grow under almost balanced conditions.

4 Conclusion

The TiO2 content in the smelting titanium slag is 46.8 %. The common minerals found in the materials include anosovite, spinel, silicate minerals, and glass phase. At a fast cooling rate, the mineral crystal size in the smelting titanium slag is small and exhibits incomplete development.

Cooling rate significantly affects the crystallization of anosovite. Slow cooling is conducive to crystal growth and improves crystal volume and size. After slow cooling treatment is performed, anosovite and gangue minerals in the smelting titanium slag reconstruct and elements redistribute. The size of anosovite and gangue mineral crystals increases and becomes suitable for separation. The most appropriate cooling rate is 2 K·min−1.