Influence of sulfur dioxide-related interactions on PM_2.5 formation in iron ore sintering

ABSTRACT

The formation of PM_2.5 (aerosol particulate matter less than 2.5 µm in aerodynamic diameter) in association with SO₂ emission during sintering process has been studied by dividing the whole sintering process into six typical sampling stages. A low-pressure cascade impactor was used to collect PM_2.5 by automatically segregating particulates into six sizes. It was found that strong correlation existed between the emission properties of PM_2.5 and SO₂. Wet mixture layer (overwetted layer and raw mixture layer) had the function to simultaneously capture SO₂ and PM_2.5 during the early sintering stages, and released them back into flue gas mainly in the flue gas temperature-rising period. CaSO₄ crystals constituted the main SO₂-related PM_2.5 during the disappearing process of overwetted layer, which was able to form perfect individual crystals or to form particles with complex chemical compositions. Besides the existence of individual CaSO₄ crystals, mixed crystals of K₂SO₄-CaSO₄ in PM_2.5 were also found during the first half of the temperature-rising period of flue gas. The interaction between fine-grained Ca-based fluxes, potassium vapors, and SO₂ was the potential source of SO₂-related PM_2.5.

Implications: The emission property of PM_2.5 and SO₂ throughout the sintering process exhibited well similarity. This phenomenon tightened the relationship between the formation of PM_2.5 and the emission of SO₂. Through revealing the properties of SO₂-related PM_2.5 during sintering process, the potential interaction between fine-grained Ca-based fluxes, potassium vapors, and SO₂ was found to be the source of SO₂-related PM_2.5. This information can serve as the guidance to develop efficient techniques to control the formation and emission of PM_2.5 in practical sintering plants.

Introduction

Iron ore sintering is an important thermal treatment stage of the whole steel-making chain, the purpose of which is to agglomerate fine iron ores into lumpish iron ore sinter for achieving excellent metallurgical performance in blast furnace (Clout and Manuel, 2003; Remus et al., 2012). In China, sinter is the major iron-bearing burden used in blast furnace for iron making, which usually takes up a proportion exceeding 75% (Wang and Huang, 2007). However, sinter-making process is both pollution-intensive and energy-intensive; 9–12% of the total energy for the whole steel-making process is consumed in this stage. Moreover, a large amount of gaseous pollutants such as carbon oxides (CO_x), sulfur oxides (SO_x), nitrogen oxides (NO_x), etc., and aerosol particulate matter, especially PM_2.5 (aerosol particulate matter less than 2.5 µm in aerodynamic diameter), are also emitted from this process (Ooi et al., 2011; Chen et al., 2009; Menad et al., 2006).

As PM_2.5 characterizes huge specific surface area, it had been proved to act as the natural carriers of dioxins, heavy metals such as Cd and Pb, and alkaline metals such as K and Na (Shih et al., 2006, 2009; Sammut et al., 2010; Hleis et al., 2013; Tsai et al., 2007). Dioxins and heavy metals are well known as toxic compounds and can probably interfere with biochemical mechanisms (Kyotani and Iwatsuki, 2002; Copeland and Kawatra, 2011). Reducing the emission of PM_2.5 in sintering process is therefore beneficial to environmental protection and human health. However, elucidating the formation mechanism of PM_2.5 is the important basis for developing efficient techniques to control PM_2.5 emissions. Unfortunately, current research on PM_2.5 emitted from sintering stacks mainly focused on its physicochemical characteristics but afforded no information about its formation mechanism (Sammut et al., 2010; Hleis et al., 2013; Tsai et al., 2007).

In China, sintering processes are an important source of SO₂, accounting for 7% of its total industrial emissions (Wang et al., 2016). In sintering bed, formed SO₂ would undergo absorption by sintering mixture layer and then intensive release into the flue gas when flue gas temperature started to rise (Fan, 2011; Pan et al., 2011; Yu et al., 2015). During the absorbing, migrating, and releasing processes, complex physicochemical reactions are accompanied, which has the potential interaction with the formation of PM_2.5. Therefore, the detailed information and reaction mechanism about this interaction are still unknown. This investigation aims to build the linkage between the formation of PM_2.5 and the emission of SO₂. For this purpose, the integrate sintering process has been divided into six typical stages according to the changing curve of flue gas temperature for PM_2.5 sampling, and a series of techniques, including X-ray fluorescence (XRF), scanning electron microscopy–energy-dispersive X-ray spectroscopy (SEM-EDS), etc., were utilized to examine the characteristics of PM_2.5.

Materials and methods

Raw materials

For making sinter, iron ore fines, fluxes (limestone, dolomite, and quicklime), return fines (sinter with size range of <5 mm), and solid fuel are necessary raw materials. All of them were directly obtained from an integrated steelwork in China. Table 1 gives the basic chemical compositions and proportions of raw materials, based on which the final sinter product was characterized by Fe content of 56.38%, SiO₂ content of 4.92%, MgO content of 2.00%, and basicity (CaO/SiO₂, mass%) of 2.00. As can be seen from Table 1, Fe was the main component of mixed iron ores and return fines; CaO and MgO were the main components of limestone, dolomite, and quicklime; and Al₂O₃ and SiO₂ were the major components of coke breeze ash. Particularly, the total content of S in coke breeze was measured, and it was about 0.205%. About 0.16% existed in the form of organics, whereas the rest mainly existed in the form of sulfide. Besides, raw materials also contained some trace elements such as S, K, Na, and Cl.

Table 1. Chemical compositions and proportion of raw materials in mixture (mass %).

Raw materials		Mixed iron ores	Dolomite	Limestone	Quicklime	Return fines	Coke breeze
Chemical composition	Fe	60.74	0.21	0.14	0.40	56.38	0.74
	SiO2	4.65	0.87	1.31	2.85	4.92	6.20
	CaO	1.90	31.57	50.19	77.92	9.36	0.89
	MgO	1.54	19.68	3.30	3.64	2.00	0.17
	Al2O3	1.92	0.22	0.32	0.75	2.07	4.12
	S	0.016	0.015	0.015	0.070	0.009	0.205
	K2O	0.026	0.036	0.082	0.027	0.008	0.041
	Na2O	0.038	0.032	0.021	0.023	0.015	0.058
	Cl	0.015	0.015	0.014	0.011	0.007	0.015
LOI		2.64	46.91	42.27	11.29	0.00	86.70
Proportion		64.93	0.53	3.05	4.56	23.08	3.85

Note. LOI = loss on ignition at 950 °C in air.

Experimental methods

Sinter pot trials

A laboratory-scale sintering pot of depth 700 mm and diameter 180 mm was applied to simulate industrial-scale sinter-making process, which was schematically described in Figure 1. During the process, raw materials given in Table 1 were well mixed according to their proportions, with prescribed proportion of water added. The mixture was then charged into a drum for granulating fine-grained materials into granules. After that, the granules were fed into the sintering pot. After feeding, the solid fuel in the surface layer was initially ignited by a natural-gas-fueled ignition hood, and the combustion front moved downward with the support of a downdraught system. The whole sintering process started from the ignition point to the burning through point at which the flue gas had reached its maximum temperature. A K-type thermocouple was used to measure the change of flue gas temperature (FGT). For each individual sampling case, a low-pressure cascade impactor (model WY-1; Beijing Nature Environmental Science and Technology Company, Beijing, China) was used to collect particulate matter from flue gas, as shown in Figure 1. An infrared flue gas analyzer, model DELTA 65-3 (MRU Corporation, Heilbronn, Germany), was used to analyze the concentration of SO₂ in the flue gas.

Figure 1. Schematic diagram for PM sampling from sintering flue gas.

PM sampling and analytical techniques

Particulate matter (PM) samples were collected by Whatman QMA quartz fiber filters (HF; Haifan Filter Material Company, Shanghai, China) and using the low-pressure cascade impactor, which was able to segregate PM into 0.7–2.5 μm and 2.5–10 μm according to their aerodynamic diameter. The part of PM with size range 0.7–2.5 μm was selected and analyzed to reflect the properties of PM_2.5. During the sampling process, water condensation inside the probe did not occur by using an electrically heated belt covering the surface of the probe. To establish the linkage between PM_2.5 formation and SO₂ emission during an integrated sintering process, it has been divided into six typical sintering stages according to the FGT curve (Figure 2). Among them, stage 1 indicated the ignition period, stage 2 showed the gradually stabilizing process of FGT, stage 3 indicated the stable process of FGT, stage 4 was the short period before the rise of FGT, and stage 5 and stage 6 constituted the first half and the second half of FGT-rising process, respectively.

Figure 2. Typical trends of flue gas temperature in sintering process.

Relevant changing and moving characteristics of each sintering layer during typical sintering stages are schematically shown in Figure 3; the division and characteristics of each layer can be found in our previous reports (Yu et al., 2016). Figure 3a corresponds to the ignition process, where combustion layer and drying & preheating layer formed gradually in this stage; Figure 3b indicates the gradual forming process of overwetted layer after the ignition process; Figure 3c indicates the typical sintering process with distinguishable five layers; Figure 3d shows the gradual disappearing process of overwetted layer; Figure 3e and 3f respectively describe the gradual disappearing process of drying & preheating layer and combustion layer, where the sintering process is to reach the burning through point. PM samples of each stage were collected respectively by repeating the sintering process with the same condition.

Figure 3. Typical changes of sinter layers during sintering process. (a) Gradual forming process of combustion layer and drying & preheating layer; (b) gradual forming process of overwetted layer; (c) formed overwetted layer; (d) gradual disappearing process of overwetted layer; (e) gradual disappearing process of drying & preheating layer; (f) gradual disappearing process of combustion layer.

For guaranteeing the representatives of PM samples, isokinetic sampling is the prerequisite. To achieve this goal, the real-time linear velocity of flue gas is calculated on the basis of real-time flue gas flow and the dimension of the flue. This information serves as the guidance to adjust the linear velocity of sampled gas, thereby making linear velocity of flue gas and sampled gas equal. After sampling, the upper part of particles with size range 0.7–2.5 μm was removed from the filter for further analysis. The elemental composition of PM_2.5 was analyzed by energy-dispersive XRF (X-ray fluorescence; Eagle III; EDAX, Inc., Mahwah, NJ, USA). Field emission scanning electronic microscopy (FESEM; Helios NanoLab 600i) with energy-dispersive X-ray (EDX) was used to analyze particle morphology and major chemical distribution in typical particles.

Results and discussion

Linkage of emission properties between PM_2.5 and SO₂

The emission properties of PM_2.5 and SO₂ during sintering process are given in Figure 4. Obviously, it was found that the emission concentration of PM_2.5 kept at a lower level from stage 1 to stage 3, whereas its emission concentration become considerably high in stage 4 and stage 5. This emission property showed well similarity to that of SO₂. As could be observed from Figure 4, the average emission concentration of SO₂ exhibited a lower value from stage 1 to stage 4, whereas it reached the peak value in stage 5. Previous research (Pan et al., 2011) had proved that wet mixture layer (raw mixture layer and overwetted layer), especially the overwetted layer, was able to capture SO₂ from flue gas via gas-solid and gas-liquid reactions, whereas SO₂ was released back into flue gas when the high temperature was approached. In combination with the change of typical sintering layers during stage 1 to stage 4 shown in Figure 3, the absorption effect of wet layers caused the lower emission concentration of SO₂. When the sintering process moved to stage 5, the temperature of sintering bed started to rise, which therefore led to desorption of SO₂.

Figure 4. Emission property of SO₂ and PM_2.5 during sintering process.

As for the emission property of PM_2.5, it was believed that wet layer has the ability to store particulate matter temporarily. Indeed, previous research (Khosa et al., 2003; Nakano and Okazaki, 2007; Debrincat and Loo, 2007) had verified that wet mixture layer, especially the overwetted layer, played an important role in scrubbing dust from flue gas. However, their research gave no information about the function of wet mixture layer on scrubbing finer particulate matter such as PM_2.5. From sintering stage 1 to stage 3, the wet mixture layer was relevantly thicker compared with that in stage 4. More PM_2.5 could be captured by the wet mixture layer due to strong capture function, which therefore showed lower emission concentration in flue gas. Sintering stage 4 referred to the gradual disappearing process of wet mixture layer. This function was correspondingly weakened in this stage. Therefore, the captured PM_2.5 was able to escape from the sintering bed, and the newly formed PM_2.5 could also easily pass through the sintering bed due to the weak function. This was the main reason for the low emission concentration of PM_2.5 from stage 1 to stage 3. More information about the function mechanism of wet mixture can also be found in our previous research. (Gan et al., 2015)

Additionally, the content of S was relatively lower in PM_2.5 from stage 1 to stage 3, whereas its content reached the highest level in PM_2.5 from stage 5, which showed well relativity to the emission concentration of SO₂ during different sintering stages. This interesting phenomenon potentially indicated that SO₂ served as an important participant in PM_2.5 formation, especially in the formation of PM_2.5 emitted from stage 4 and stage 5, where absorbed SO₂ started to release back into the flue gas.

Characteristics of SO₂-related PM_2.5 from stage 4 and stage 5

SO₂-related PM_2.5 from stage 4

During sintering stage 4, overwetted layer began to disappear gradually, and scrubbed SO₂ and PM_2.5 started to transfer from sintering bed to flue gas. The morphology and major inorganic composition of SO₂-derived PM_2.5 are given in Figure 5 and Figure 6.

Figure 5. SEM-EDS of perfect crystallized and nonperfect crystallized CaSO₄ crystals. (a) Morphology of perfect crystallized CaSO₄ crystals; (b) EDX spectrum of perfect crystallized CaSO₄ crystals; (c) morphology of nonperfect crystallized CaSO₄ crystals; (d) EDX spectrum of nonperfect crystallized CaSO₄ crystals.

Figure 6. SEM-EDS of CaSO₄-assembled particles. (a) EDX spectrum of selected area 1; (b) EDX spectrum of selected area 2.

As could be seen from Figure 5a and 5b, the particles with smooth surface and standard polyhedral structure mainly consisted of Ca, S, and O, which accounted for 24.74%, 22.02%, and 44.43% of the total weight, respectively. It could be found that the molar ratio of Ca, S, and O was about 1:1:4. Therefore, this kind of particles was the typical CaSO₄ crystals. Additionally, the rod-like particle shown in Figure 5c also mainly comprised Ca, S, and O. However, Fe took up about 10% of the total mass in this particle, as shown in Figure 5d. Taking the molar ratio of Ca and S, which is about 1:1, into consideration, this particle was a nonperfect crystallized CaSO₄ crystal. Moreover, taking the main chemical composition of this particle into account, Fe was probably existed in the form of oxides or sulfates.

Figure 6 gives the morphology of an irregularly shaped particle assembled with numerous finer-grained particles. As for the area 1, the main chemical composition was O, Si, Al, and Ca, which accounted for about 43.29%, 24.78%, 9.12%, and 10.21% of the total mass, respectively. It could be speculated that this part of particle was the mixture of SiO₂, Al₂O₃, and CaO. Area 2 consisted of many particles less than 1.0 μm in diameter, the major chemical composition of which was Ca, S, and O, with mass proportion of 25.05%, 19.77%, and 42.90%, respectively. Also, it was found that the particle was a CaSO₄ crystal according to the molar ratio of Ca, S, and O. Consequently, CaSO₄ was able to form submicron particles and assemble together to form larger particles.

SO₂-related PM_2.5 from stage 5

During sintering stage 5, FGT started to rise, which was accompanied by the large-scale release of SO₂ and PM_2.5 into flue gas. The properties of SO₂-related particles from this stage are given in Figures 7, 8, and 9.

Figure 7. SEM-EDS of CaSO₄-K₂SO₄ hybrid crystals.

Figure 8. SEM-EDS mapping of CaSO₄-K₂SO₄ hybrid particles.

Figure 9. SEM-EDS mapping of CaSO₄ particles.

As could be observed in Figure 7, Ca, S, O, K, and Cl took up the major part of the total mass in the rod-like particle. The high K content was in great agreement with the high K contents of PM_2.5 in this stage, as reported in our previous research (Gan et al., 2015). According to the mass ratio of S and O, it was observed that their molar ratio was about 1:4. Therefore, S and O existed in this particle in the form of sulfate. Taking the mass proportion of K and Ca into consideration, it could be found that the molar ratio of K, Ca, and S was about 2:1:2, which indicated that the rod-like particle was the hybrid crystals of K₂SO₄ and CaSO₄, with molar ratio 1:1. The SEM-EDS mapping of the rod-like particle was described in Figure 8, where it could be seen clearly that the distribution areas of K, Ca, S, and O were well overlapped, which further verified that K₂SO₄ and CaSO₄ participated in the formation of PM_2.5 simultaneously. From Figure 8, it was also found that K could participate in the formation of PM_2.5 by forming KCl. Consequently, the contribution of K₂SO₄ and KCl during PM_2.5 formation possibly existed some competitiveness.

With the exception of forming mixed crystal with K₂SO₄, CaSO₄ could also participate in forming PM_2.5 individually. Figure 9 gives the SEM-EDS mapping of a polyhedral particle. On the basis of the contents and distribution property of Ca, S, and O in this particle, it could be observed that this particle was also CaSO₄.

Discussion on the formation of SO₂-related PM_2.5

The source of CaO for the formation of CaSO₄ in PM_2.5

According to the results shown in “Characteristics of SO₂-related PM_2.5 from stage 4 and stage 5,” CaSO₄ crystals were the important SO₂-related PM_2.5 formed from both stage 4 and stage 5. Except for the participating of SO₂, Ca-based compounds were also necessary in the formation of CaSO₄ crystals. As introduced in Table 1, three kinds of Ca-based fluxes were used as the necessary materials for producing sinter, which could provide abundant CaO to capture SO₂ from flue gas. However, the size distribution of Ca-based fluxes served as another important factor to decide the dimension of the SO₂-related particulate matter formed. For elucidating the possibility of the transformation of gaseous SO₂ into PM_2.5 in the form of CaSO₄ crystals, the size distributions of slaked lime, dolomite, and lime stone were measured by laser diffraction using a Malvern Mastersizer 3000, as shown in Figure 10. Obviously, all of Ca-based fluxes contained a part of particles less than 2.5 µm in diameter, which therefore guaranteed the appearance of CaSO₄ in PM_2.5 via the main reactions given in eqs 1–7, as reported previously (Pan et al., 2011; Yu et al., 2015; Li et al., 2014):

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Figure 10. Size distributions of basic fluxes (<0.5 mm).

The source of K for the formation of K₂SO₄ in PM_2.5

K was the trace elements in raw materials as introduced in Table 1, and it was able to volatilize partially in the sintering process, where the highest temperature easily reached above 1300 °C, as has been reported (Remus et al., 2012; Yu et al., 2012). Actually, the removal ratio of K during the sintering process with increasing ratio of coke breeze had been investigated. From the results shown in Table 2, it could be observed that K achieved an increasing removal ratio with enhancing the ratio of coke breeze. Reaction 8 could serve as the explanation to the increased removal ratio of K (Zhang et al., 2010), which showed that K can enter into the flue gas in the form of its vapors. As has been widely known, K is of extremely high chemical activity and could react with O₂, CO₂, SO₂, etc., in the flue gas. Among these, the reactions 10 and 11 potentially constituted the main routes of the formation of K₂SO₄.

(8)

(9)

(10)

(11)

Table 2. Removal ratio of K with increasing the ratio of coke breeze.

Ratio of coke breeze (%)	Content of K in raw materials (mg/kg)	Mass of raw materials (kg)	Content of K in sinter (mg/kg)	Mass of sinter (kg)	Removal ratio of K (%)
4.70	699.15	1	653.11	0.833	16.66
5.00	734.42	1	626.11	0.799	23.94
5.30	746.63	1	588.97	0.751	29.63
5.60	667.00	1	473.94	0.605	36.61

Formation mechanism of SO₂-related PM_2.5

During the sintering process, water vapor generated from the drying & preheating layer would condense on the surface of granules in the lower material layer, thereby forming the overwetted layer. Results introduced in “Linkage of emission properties between PM_2.5 and SO₂” had clarified that wet mixture layer, especially the overwetted layer, had the ability to capture PM_2.5 and SO₂ from flue gas, which occurred from sintering stage 1 to stage 3, as shown in Figure 11. During the absorption process, Ca-based basic fluxes with size range less than 2.5 µm could serve as the efficient absorbent to absorb SO₂, thereby forming PM_2.5 in the form of CaSO₄ crystals. Besides, in combination with the distinguishingly high K content in PM_2.5 from stage 5 (Gan et al., 2015), K-bearing compounds were first captured in the wet mixture layer and then intensively released into the flue gas, as also found elsewhere (Kobayashi et al., 1979). Therefore, the potassium vapor generated from the reduction process of its compounds also participated in the formation of PM_2.5, which could be transformed into new compounds such as KCl and K₂SO₄, the existence of which had been verified, with SO₂, O₂, CaCl₂, etc., and were temporarily captured in the wet mixture layer. As the absorption reactions occurred on the surface of the granules, generated K₂SO₄ had the convenience to collide with PM_2.5 in the form of CaSO₃ or CaSO₄ crystals, and finally forming CaSO₄-K₂SO₄ mixed crystals through particle aggregation. When the sintering process moved to stage 4 and stage 5, wet mixture layer started to disappear. PM_2.5 was correspondingly rereleased into the flue gas. SO₂-related PM_2.5 such as CaSO₄ and CaSO₄-K₂SO₄ crystals were also released into flue gas (Figure 11).

Figure 11. Proposed formation and release mechanism of SO₂-related PM_2.5.

Conclusion

The formation of PM_2.5 in association with SO₂ during sintering process has been studied by dividing the whole sintering process into six typical sampling stages. On the basis of the properties of the SO₂-related PM_2.5, conclusions can be drawn as follows.

1. Great similarity is existed between the emission properties of PM_2.5 and SO₂. Wet mixture layer has the same function to capture SO₂ and PM_2.5 during the early sintering stages, and to release them back into flue gas mainly in the temperature-rising period of flue gas.

2. CaSO₄ crystals are the main SO₂-related PM_2.5 during the disappearing process of overwetted layer, which is able to form perfect individual crystals or particles with complex chemical composition. Besides the existence of individual CaSO₄ crystals, mixed crystal of K₂SO₄-CaSO₄ is also observed during the first half of the rising period of flue gas temperature.

3. The interactions between fine-grained basic fluxes, potassium vapors, and SO₂ in the wet mixture layer serve as the potential sources of SO₂-related PM_2.5.

Funding

The authors are grateful to the National Natural Science Foundation of China (51304245), the Postdoctoral Science Foundation (2013M540639), Hunan Provincial Co-Innovation Center for Clean, and Efficient Utilization of Strategic Metal Mineral Resources, and the Fundamental Research Funds for the Central Universities of Central South University (2015zzts086) for supporting this research.

Additional information

Funding

The authors are grateful to the National Natural Science Foundation of China (51304245), the Postdoctoral Science Foundation (2013M540639), Hunan Provincial Co-Innovation Center for Clean, and Efficient Utilization of Strategic Metal Mineral Resources, and the Fundamental Research Funds for the Central Universities of Central South University (2015zzts086) for supporting this research.

Notes on contributors

Zhiyun Ji

Zhiyun Ji is a Ph.D. at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Xiaohui Fan

Xiaohui Fan is a research professor at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Min Gan

Min Gan and Xuling Chen are research professors at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Xuling Chen

Min Gan and Xuling Chen are research professors at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Wei Lv

Wei Lv, Qiang Li, Yang Zhou, and Ye Tian are postgraduate fellows at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Qiang Li

Wei Lv, Qiang Li, Yang Zhou, and Ye Tian are postgraduate fellows at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Yang Zhou

Wei Lv, Qiang Li, Yang Zhou, and Ye Tian are postgraduate fellows at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Ye Tian

Wei Lv, Qiang Li, Yang Zhou, and Ye Tian are postgraduate fellows at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.

Tao Jiang

Tao Jiang is a research professor at the Department of Iron and Steel Metallurgy, Central South University, in Changsha, China.