1 Introduction

Iron ore is the most important material for iron and steel production [27, 28]. However, with the depletion of high-grade hematite and magnetite ores, the utilization of low-grade iron ore gradually caused more attention [9]. In China, most of the iron resources are refractory iron ore, which is characterized by complex mineral composition, low iron grade, and fine dissemination [8]. Limonite ore is typical refractory iron ore and is widely distributed in Australia, Brazil, and China. Recently, the Pilbara region of Western Australia is the largest producer of limonite ores with more than 130 million tonnes per year [6]. In China, the reverse of limonite is more than 1.23 billion tons. Additionally, most refractory iron ore has an amount of limonite in mineral composition [7]. Thus, the utilization of limonite ore resources is important for the iron and steel industry.

Limonite ore is a compound that consisted of iron oxides, iron hydroxides, aqueous quartz, and argillaceous minerals. The primary iron minerals are hematite, goethite, and lepidocrocite [17, 26]. Therefore, limonite ore is usually characterized by low iron grade, high crystal water, and fine dissemination. Owing to the weak magnetism and easy sliming in the crushing and grinding process, limonite ore is difficult to process using conventional methods including gravity concentration, magnetic separation, and froth flotation [24, 25]. In recent years, new processing methods like high-intensity magnetic separation, flocculation magnetic separation, flotation, leaching, and their combinations have been reported [5, 10, 15, 35], but their industrial application has not been achieved due to the complex flowsheet and unsatisfied separation index. The typical Xinyu limonite ore was treated using a combined process of magnetic separation and flotation, and an iron concentrate with grade of 54% and recovery of 60% was obtained.

Magnetization roasting is an effective method to treat limonite ore [19, 21, 24], in which weakly magnetic hematite, goethite, and lepidocrocite were reduced into ferrimagnetic magnetite [33, 37], and then the iron can be recovered by low-intensity magnetic separation. Despite this, the application of magnetization roasting in the utilization of limonite ore has not been widely achieved. The reason is that the available magnetization roasting equipment such as shaft furnaces and rotary kilns have the disadvantages of high coal dosage, uneven reduction, and low operation efficiency, which can not meet the clean and efficient requirements of industrial production [29].

$$\mathrm{\alpha }-\mathrm{FeOOH}\to \mathrm{\alpha }-{\mathrm{Fe}}_{2}{\mathrm{O}}_{3}+{\mathrm{H}}_{2}\mathrm{O}$$
(1)
$$\upgamma -\mathrm{FeOOH}\to \mathrm{\alpha }-{\mathrm{Fe}}_{2}{\mathrm{O}}_{3}+{\mathrm{H}}_{2}\mathrm{O}$$
(2)
$$\mathrm{\alpha }-{\mathrm{Fe}}_{2}{\mathrm{O}}_{3}+\mathrm{CO}\to {\mathrm{Fe}}_{3}{\mathrm{O}}_{4}+\mathrm{CO}_{2}$$
(3)
$$\mathrm{\alpha }-{\mathrm{Fe}}_{2}{\mathrm{O}}_{3}+{\mathrm{H}}_{2}\to {\mathrm{Fe}}_{3}{\mathrm{O}}_{4}+{\mathrm{H}}_{2}\mathrm{O}$$
(4)

Innovative technology and equipment have been reported with the development of magnetization roasting technology [14, 34, 39]. A multi-stage fluidization roasting becomes a promising method to treat refractory iron ore such as hematite, limonite, and siderite [19, 20, 32]. Particularly, the multi-stage reactors are effective to treat limonite ore because iron oxide hydroxides are first dehydroxylated in the pre-heating reactor and then are reduced to magnetite in the reduction reactor [36]. In contrast, there is only one reduction reactor in conventional roasting equipment such as shaft furnaces and rotary kilns,the complex dehydroxylation and reduction reactions (Eqs. 1–4) occurred synchronously in the reduction reactor. It usually causes the over-reduction and incomplete reduction of limonite ore in magnetization roasting and lowers the iron recovery in subsequent magnetic separation. Thus, the key to the magnetization roasting of limonite ore is the precise control of phase transformation.

Although relevant references have reported the reduction roasting of limonite ore in CO or H2 atmospheres, the impact of dehydroxylation has not been focused on. The dehydroxylated characteristics exhibited great differences owing to the complex mineral composition of limonite ore. In this study, a limonite ore obtained from Yunnan, China, was used as the material. The samples were analyzed using chemical elemental analysis, in situ X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FT-IR), and simultaneous thermogravimetry and differential scanning calorimetry (TG-DSC) analysis. The phase transformation and reaction kinetics were conducted in the dehydroxylation of limonite ore.

2 Experimental

2.1 Materials

The limonite ore was obtained from Yunnan, China. The chemical composition analysis is shown in Table 1. The iron content was 54.58%. The contents of SiO2, CaO, MgO, and Al2O3 were 6.56%, 0.18%, 0.15%, and 0.30%, respectively. In addition, the contents of Cu, Pb, and Mn were 0.38%, 1.27%, and 0.52% respectively. Meanwhile, the content of harmful element S was 0.07%. The loss on ignition (LOI) was 10.97%. XRD analysis (Fig. 1) indicated that the primary minerals were goethite, hematite, lepidocrocite, and quartz. The particle size result (Fig. 2) found that 90% (volume content) of the sample was smaller than 179.30 μm.

Table 1 Chemical composition analysis of limonite ore (mass, %)
Full size table
Fig. 1
figure 1

XRD analysis of limonite ore

Full size image
Fig. 2
figure 2

Particle size analysis of the limonite ore

Full size image

2.2 Experimental Method

The chemical elemental analyses of limonite ore were conducted; the Fe and FeO were determined using chemical titration method; the SiO2, Al2O3, MgO, CaO, Cu, Pb, Mn, and P were determined using X-ray fluorescence (XRF) spectrometry (Primus II, Japan); the S was determined using a LECO SC-144DR sulfur analyzer. The particle size was analyzed using laser diffraction particle size analysis (Malvern Panalytical Mastersizer 3000, England).

2.3 Characterization

The dehydration thermal decomposition study of limonite samples was carried out by non-isothermal thermal analysis using a comprehensive thermal analyzer (NETZSCH STA 409 PC/PG, Hamburg, Germany). Each 30 mg sample was used for detection at the heating rate of 5, 10, 15, 20, and 25 °C/min, and nitrogen (99.99%) was used as the protective gas in the detection process. The test temperature range of each sample was from 30 to 600 °C. The thermogravimetry (TG) and derivative thermogravimetry (DTG) analysis results were obtained by the software NETZSCH Proteus Thermal Analysis (6.0.1d).

The mineralogical composition of limonite samples was detected by a polycrystalline X-ray diffractometer (PW3040, PANalytical B.V., Netherlands). The polycrystalline X-ray diffractometer was equipped with a 2.2 kW Cu anode with a long, fine-focus ceramic X-ray tube for generating Cu Ka radiation. Scanning range during detection was 5–90°; scanning speed was 12°/min.

The phase transformation of products at different temperatures was detected by an in situ polycrystalline X-ray diffractometer (Empyrean Malvern Panalytical B.V., Netherlands), equipped with an Anton Paar HTK 1200 N oven-chamber. In high-temperature XRD detections, the limonite ore sample was heated in Anton Paar HTK 1200 N oven-chamber with a heating rate of 10 °C/min from 20 to 500 °C. The scanning range was 5–90°, and the scanning speed was 12°/min. The XRD patterns were analyzed using the software HighScore Plus (3.0e).

A Thermo Scientific FT-IR analyzer (Nicolet iS5, Thermo Fisher Scientific, Waltham, MA, USA) was used to analyze the structural changes of limonite at different roasting temperatures. Firstly, 6 g of limonite samples were roasted in a small vertical tube furnace (OTF-1200X-S-VT, HF-Kejing Hefei China) [38] at different temperatures for 12 min, and then the obtained roasted products were used as test samples for detection. The samples to be tested were mixed with potassium bromide at a ratio of 1:150 at 20 °C and compressed into tablets. The spectra were obtained in the scanning range between 400 and 4000 cm−1, and the number of scans was 32. The atmospheric background (H2O and CO2) was deducted when processing spectral data. Some schematic diagrams of the characterization methods are shown in Fig. 3.

Fig. 3
figure 3

The schematic diagram of the laboratory fluidization roasting reactor and characterization methods for limonite ore dehydroxylation process

Full size image

2.4 Kinetics Analysis

Thermogravimetric (TG) analysis of samples at different heating rates is the main method used in the study of non-isothermal kinetics [22, 30]. The raw samples were heated from 30 to 600 °C at the heating rate of 5, 10, 15, 20, and 25 °C/min, protected in N2 with a flow rate of 150 mL·min−1. Conversion fraction α was calculated according to Eq. (5) using TG analysis data:

$$a=\frac{{m}_{0}-{m}_{t}}{{m}_{0}-{m}_{\infty }}$$
(5)

where m0 is the initial mass of limonite ore before dehydroxylation, mt is the sample mass at a particular time (g), and m∞ is the theoretical mass (g) of limonite ore after dehydroxylation process.

The calculation method of the reaction rate r during the thermal decomposition reaction of limonite is shown in Eq. (6):

$$r=\frac{d\alpha }{dt}=\frac{\Delta \alpha }{\Delta t}$$
(6)

where ∆α is the weight loss in a short span of time (g) and ∆t is the short time interval (min).

In this experiment, the reaction rate of a heterogeneous reaction under non-isothermal conditions with a constant heating rate can usually be described by Eq. (7) [22]:

$$\frac{d\alpha}{dt}=\beta\frac{d\alpha}{dT}=A\;\exp\left(\frac{-E}{RT}\right)f(\alpha)$$
(7)

where β is the heating rate, °C·min−1, E is the activation energy (kJ·mol−1), A is the pre-exponential factor (min−1), T is the temperature (Kelvin, K), t is the time (min), R is the gas constant (8.314 J·mol−1·K−1), and f(α) is the differential conversion function.

The integral form of the function is obtained by permuting and integrating Eq. (7), as shown in Eq. (8):

$$G(\alpha )={\int }_{0}^{\alpha }\frac{d(\alpha )}{f(\alpha )}=\frac{A}{\beta }{\int }_{{T}_{0}}^{T}\mathrm{exp}(\frac{-E}{RT})dT$$
(8)

Numerous analytical methods have been proposed to obtain the non-isothermal kinetic parameters, among which the Šatava–Šesták integral method is usually used to determine the kinetic mechanism function of non-isothermal reactions, as shown in Eq. (9):

$$\lg\;G(\alpha)=\lg\left(\frac{AE_s}{R\beta}\right)-2.315-0.4567\frac{E_s}{RT}$$
(9)

Ozawa–Flynn–Wall (OFW) method is often used to calculate the activation energy of the reaction [2, 31]. It was proposed by Flynn, Wall, and Ozawa, and its form is shown in Eq. (10). In this study, the Ozawa–Flynn–Wall method is used to calculate the activation energy of the non-isothermal decomposition of limonite ore.

$$\mathrm{lg}(\beta )=\mathrm{lg}\left(\frac{AE}{RG(\alpha )}\right)-2.315-0.4567\frac{E}{RT}$$
(10)

The optimal mechanism function can be determined by analyzing the linear relationship between G(a) and 1/T. In this paper, 30 commonly used kinetic mechanism functions (Table 2) were used to fit and analyze the non-isothermal experimental data [9, 13, 18]. Ea, lnA, and the correlation coefficient R were calculated using the least squares method for G(a) and 1/T linear fit. The optimal mechanism function was determined based on the linear relationship. Additionally, the activation energy was also calculated by the Ozawa–Flynn–Wall method to verify the best mechanism function. The kinetic model of the reaction can be established according to the determined mechanism function.

Table 2 Differential and integral expressions of common reaction mechanism functions
Full size table

3 Results and Discussion

3.1 In Situ XRD Analysis

The dehydroxylation of limonite ore was analyzed via in situ XRD, and the result is shown in Fig. 4. When the temperature was lower than 250 °C, characteristic diffraction peaks did not change significantly, and goethite still existed. Although Liu et al. [12] reported that a natural goethite ore was transformed into hematite at 250 °C, the phase transformation from goethite to hematite was very slow below 300 °C in this study based on diffraction peaks in Fig. 4. After the temperature reached 300 °C, goethite characteristic peaks gradually weakened and disappeared. Meanwhile, the hematite characteristic peaks became broad and strengthened. This is because goethite dehydroxylated into hematite in this process. The characteristic peak variation is consistent with the Gialanella et al. [4] study, in which the initial temperature of transformation from goethite to hematite was 295 °C. The characteristic peaks of goethite vanished at the temperature of 350 °C, indicating the dehydroxylation was complete. With further increase in temperature, hematite characteristic peaks gradually strengthened, proving that the synthetic hematite crystal structure tended to be perfect.

Fig. 4
figure 4

The in situ XRD analysis of limonite ore at temperatures of 20–500 °C

Full size image

3.2 FT-IR Analysis

The limonite ore and its dehydroxylated samples at temperatures of 300, 400, 500, and 600 °C were analyzed using FT-IR, and the results are shown in Fig. 5. Based on the FT-IR spectrum of limonite ore, a typical signal for goethite and lepidocrocite phases was an intense band appeared at 3134.29 cm−1, which corresponded to the absorption peaks of -OH in α-FeOOH. The absorption peaks at 1633.79 cm−1 corresponded to the bending vibration absorption peaks of H2O [3]. The absorption peak at 895.20 and 797.51 cm−1 was attributed to the stretching vibration absorption peak of δ(-OH) and γ(-OH) hydroxyl functional groups, respectively [1]. The absorption peak at 597.83 cm−1 was attributed to the bending vibration of -OH. The absorption peak at 458.56 cm−1 was the stretching vibration absorption peak of Fe–O in α-FeOOH. When temperature exceeded 300 ℃, the characteristic peaks at 895.20 and 797.51 cm−1 bands significantly weakened, and peaks at 541.88 and 461.89 cm−1 presented in the low-frequency region. This is attributed to the stretching vibration of Fe–O in the transformation from goethite to hematite. Meanwhile, in the high-frequency region of 3401 and 3134 cm−1, the vibrational characteristic peaks of hydroxyl groups disappeared, and the H2O-band between 3375 and 3430 cm−1 of surface-adsorbed water appeared. With the increase in temperature, the peaks at 895.20 and 797.51 cm−1 gradually vanished, indicating the dehydroxylation become complete. After dehydroxylation, peaks of hydroxyl and water molecules at the 3401 and 3134 cm−1, 1516.37, and 1633.79 cm−1 bands should be significantly reduced. However, owing to the limitation of experimental conditions, the sample was inevitably exposed to the air during the measurement, so the sample recombined with the adsorbed water, resulting in their appearance in the measured FT-IR spectrum.

Fig. 5
figure 5

FT-IR spectra of limonite ore and roasted samples at 300, 400, 500, and 600 ℃ for 10.0 min

Full size image

3.3 TG-DTG Analysis

The TG-DTG analysis of limonite ore was conducted in N2 atmosphere with heating rates of 5, 10, 15, 20, and 25 °C/min, and the results are shown in Fig. 6. The total weight loss was 10.70% when the roasting temperature increased from 30 to 600 °C. The first weight loss of 0.40% occurred below 100 °C; the slight mass change in the range from 20 to 100 °C was also observed in synthetic and natural specimens [16] corresponding to the removal of adsorbed water in the limonite ore surface. The second weight loss was 10.30% in the range of 100 to 400 °C, which was close to the theoretical mass loss of 10.11% in the dehydroxylation reaction of goethite and lepidocrocite as Eqs. (1–2). Liu et al. [11] also reported that the second weight loss for synthetic goethite was 10.1–10.7%. With the increase in heating rate from 5 to 25 °C/min, the peak temperature varied from 310.33 to 343.04 °C. Meanwhile, the mass loss rate increased from 0.69 to 2.96%/min−1.

Fig. 6
figure 6

TG-DTG analysis of limonite ore in N2 atmosphere (a) TG curves and (b) DTG curves

Full size image

3.4 Kinetics Analysis

Based on the thermogravimetric analysis data at heating rates of 5, 10, 15, 20, and 25 °C/min, the reaction fraction(a) and reaction rate (r) of the dehydroxylation were calculated, and the results are shown in Fig. 7 and Fig. 8, respectively. The dehydration of limonite ore was primarily divided into three parts. The reaction fraction increased slowly between 0 and 0.15 in the initial stage. Then it drastically increased from 0.15 to 0.90 in the second stage. Finally, the reaction fraction slowly increased to 1.0. Figure 8 exhibited the reaction rate curves for different heating rate conditions between 300 and 350 °C. With the increase in heating rate, the maximum reaction rate increased from 0.063 to 0.289 min−1, and the dehydroxylation time significantly shortened.

Fig. 7
figure 7

Reaction fraction at different heating rates

Full size image
Fig. 8
figure 8

Reaction rate at different heating rates

Full size image

The initial stage of the reaction is mainly the removal of adsorbed water inside the solid, while the solid reaction is unstable in the initial stage and the final stage. Additionally, the dehydroxylation of limonite ore primarily occurred in the second stage; the kinetic analysis was calculated based on the reaction fraction of 0.15–0.90. The fitting of lg(β) and 1/T using the Ozawa–Flynn–Wall method was calculated based on Eq. (10), and the results are shown in Figs. 9 and 10. There was a good linear relationship between lg(β) and 1/T, and the calculated activation energy was 113.94 kJ mol−1.

Fig. 9
figure 9

Regression lines for various degree of reaction based on OFW method

Full size image
Fig. 10
figure 10

Dependence of the activation energy on the degree of reaction fraction

Full size image

Based on the Šatava–Šesták integral method, the kinetic data is substituted into the kinetic mechanism function of the reaction, some optimal kinetic function models with correlation coefficients R2 > 99% were determined, the activation energy was calculated, and the results are shown in Table 3. The activation energy of the 2-D diffusion (Jander) reaction model is roughly equal to the Ea estimated by the Ozawa–Flynn–Wall method (Fig. 10). Hence, the 2-D diffusion (Jander) reaction model, G(α) = [1 − (1 − α)1/2]2 and f(α) = (1 − α)1/2[1 − (1 − α)1/2]−1, was selected as the most probable mechanism function. The linear fitting of G(a) versus 1/T based on the 2-D diffusion (Jander) model was conducted, shown in Fig. 11. The calculated activation energy and pre-exponential factor were 113.43 kJ·mol−1 and 8.50 min−1, respectively. Compared with previous studies, it was found that the dehydroxylation mechanism has significant differences owing to the microstructure, impurity ion substitution, porosity, mineral composition, and geological origin of limonite ore. Liu et al. [11] investigated the effect of aluminum substitution on goethite dehydration, and the calculated activation energy for pure synthetic goethite and Al-substituted goethite were 75.5 and 94.8 kJ·mol−1, respectively. Walter et al. [23] studied the effect of particle size on dehydroxylation of limonite, the activation energies ranged from 107 to 138 kJ·mol−1. In this study, the limonite ore primarily consisted of goethite and lepidocrocite, and some Fe ions were substituted with Pb ions in the lattice structure, which also affected the calculated activation energy.

Table 3 Optimal mechanism function results calculated by Šatava–Šesták integral method
Full size table
Fig. 11
figure 11

Linear fitting of G(a) versus 1/T based on the 2-D diffusion reaction model

Full size image

4 Conclusion

The dehydroxylation mechanism of a natural limonite ore was conducted to study the phase transformation and kinetics. The dehydroxylation of limonite ore (goethite and lepidocrocite) primarily began at the temperature of 300 °C and was complete at the temperature350 °C based on the in situ XRD analysis. In this dehydroxylation, the stretching vibration absorption peak of δ(-OH) and γ(-OH) at 895.20 and 797.51 cm−1 significantly weakened and disappeared when the temperature exceeded 300 °C. Based on the thermogravimetric analysis from 30 to 600 °C with heating rates of 5, 10, 15, 20, and 25 °C/min in the N2 flow atmosphere, the dehydroxylation reaction model of limonite ore was calculated. The dehydroxylation kinetic followed the 2-D diffusion (Jander) reaction model G(α) = [1 − (1 − α)1/2]2, and the calculated activation energy and pre-exponential factor were 113.43 kJ·mol−1 and 8.50 min−1, respectively. It has a good agreement with the Ea calculation of 113.94 kJ mol−1 using the Ozawa–Flynn–Wall method. This study is conducive to the utilize limonite ore resource using a multi-stage fluidization roasting technology.