A novel Ti-based sorbent for reducing ultrafine particulate matter formation during coal combustion

A novel Ti-based sorbent for reducing ultrafine particulate matter formation during coal combustion

Fuel 193 (2017) 72–80 Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel Full Length Article A novel Ti...

2MB Sizes 0 Downloads 116 Views

Fuel 193 (2017) 72–80

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

A novel Ti-based sorbent for reducing ultrafine particulate matter formation during coal combustion Yishu Xu, Xiaowei Liu ⇑, Yu Zhang, Wei Sun, Yingchao Hu, Minghou Xu ⇑ State Key Laboratory of Coal Combustion, School of Energy and Power Engineering, Huazhong University of Science and Technology, Luoyu Road 1037, Wuhan 430074, China

a r t i c l e

i n f o

Article history: Received 4 July 2016 Received in revised form 9 December 2016 Accepted 15 December 2016

Keywords: Sorbent Ultrafine particulate matter Coal combustion In-furnace Additive

a b s t r a c t Ultrafine particulate matter (PM) is an important part of PM2.5, which is enriched with hazardous components and more harmful; meanwhile it cannot be effectively removed by the common-used dust collectors. In-furnace sorbent injection is an emerging technology to reduce the emission of PM in the coal combustion and the sorbent is a key factor determining its feasibility. In this study, to seek new PM sorbents, eight minerals were first separately added into a pulverized coal and burned in a drop-tube furnace (DTF) at 1773 K. The derived PM was collected via a Dekati Low pressure impactor (DLPI) sampling system and impacts of each mineral on the PM emission were evaluated. Then, the tested minerals were burned with pure sodium acetate (NaAc) under the same conditions to determine their Na fixation abilities. Finally, the PM reduction mechanism of the sorbent was discussed based on the particle size distribution, mass yield, composition and micromorphology of PM and the chemical and physical properties of the sorbent. A novel Ti-based PM reduction sorbent was screened out, which exhibited an ultrafine PM (PM0.2) reduction efficiency of 39% under the experimental conditions at an addition ratio of 5% (wt. %, coal basis). The Ti-based sorbent reacted with Na-contained vapour and formed sodium titanates. Fixation of the Na-contained vapour by the sorbent was considered to be the primary PM capture mechanism. What’s more, the high sintering temperature of the Ti-based sorbent facilitates its PM reduction performance. Under the high temperature combustion conditions, the Ti-based sorbent exhibited a good performance in capturing ultrafine PM and Na-contained vapour, indicating its potential of being a high temperature sorbent. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Particulate matter (PM) emitted from the coal-fired power plants is causing severe ambient PM pollution in China [1,2]. Ultrafine PM (PM with the aerodynamic diameter of <0.2 lm) is of particular concerns as it is enriched with the hazardous components (e.g., trace elements, PAHs, etc.) and is much more harmful to the human health [1,3,4]. Coal combustion in the power plants is one primary PM emission source, especially the ultrafine PM [1,5]. At present, dust collectors (e.g., ESP, fabric filter) are widely installed downstream of the furnace to remove PM in the flue gas [5]. Although these dust collectors generally have total PM collection efficiency of over 99%, they cannot effectively remove the ultrafine PM [5–8]. Adding more chambers and/or electrostatic fields in the dust collectors, or improve their operation parameters may help improve the removal of fine and ultrafine PM meanwhile it is not economical or not feasible due to lacking space on the existed ⇑ Corresponding authors. E-mail addresses: [email protected] (X. Liu), [email protected] (M. Xu). http://dx.doi.org/10.1016/j.fuel.2016.12.043 0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.

power plant units. To effectively and economically improve the removal of ultrafine PM, the PM sorbent technology has been proposed, in which certain sorbents/additives are injected into the furnace and reduce the emission of PM and trace elements simultaneously during the combustion process [5,9–14]. In this way, the yield of ultrafine PM is reduced in the furnace, thereby easing the burden of the traditional PM removal devices (e.g., electrostatic precipitator, baghouse, etc.) downstream of the furnace and finally reinforcing the control of ultrafine PM. Sorbent is a key factor affecting the feasibility of this technique and developing high efficient PM sorbents has been a focus in this field. Briefly, three types of PM sorbents have been reported in public, including the Si-/Al-based, the Ca-based and the Mg-based sorbents. These sorbent materials are different in the composition and exhibit different PM removal performance and mechanism [12]. Kaolin, alumina and silica are the most widely studied Si-/Al-based sorbents [10,12,15,16]. As reported, the Si-/Al-based sorbent particles injected into the furnace would react with the mineral vapour (e.g., alkali metal, heavy metal, etc.) released from the burned coal and inhibit their migration into

Y. Xu et al. / Fuel 193 (2017) 72–80

the ultrafine PM. In this way, the formation of ultrafine PM is reduced by the so-called vapour-capture mechanism [9,17–19]. Laboratory studies by Linak et al. [9] showed that kaolin addition into the residual oil combustion reduced the ultrafine nuclei (less than 0.1 lm) by 35%. In another laboratory study on coal combustion, kaolin was reported to reduce the yield of PM1 (PM less than 1 lm) by 20–60% with an addition ratio of 5 g-kaolin/100 gcoal. Moreover, the tested kaolin exhibited the best performance at 1373 K when the combustion temperature increased from 1173 to 1573 K [10]. Our previous study further showed that kaolin (3 g-sorbent/100 g-coal) can reduce the emission of PM0.2–0.5 by 9–48% at the combustion temperature of 1773 K and the PM reduction performance of kaolin depended closely on the properties of minerals in coal and combustion atmosphere [12]. Besides, some hazardous heavy metals such as Pb, Cd and Cr are simultaneously captured by the Si-/Al-based sorbents [17–19]. The Ca-based PM sorbents mainly include limestone, hydrated lime and calcined lime [10,11]. Studies on a lab scale drop tube furnace (DTF) showed that the addition of limestone (5 g-sorbent/100 g-coal) reduced the emission of PM1 by 20–70% during the combustion of a high S-content coal at 1173–1573 K. Different from the Si-/Al-based sorbents, the Ca-based sorbents added into the furnace primarily capture the acidic species such as SOx released out of the burning coal particles. And the performance of such Ca-based sorbents is observed to be related to the S content in coal [10]. The Mg-based PM sorbents are developed by Wei and Ninomiya et al. [13,14], which mainly are Mg(OH)2 and (CH3COO)2Mg. It was reported that Mg-based sorbents exhibited removal efficiencies of 40–50% and 40–60% for PM1 and PM2.5 respectively with the coal-based addition ratio of 5%. Different from the aforementioned Si-/Al-based and Ca-based sorbents, Mg-based sorbents would enhance the melting of minerals and thereby capture the small PM through an additional liquidus-capture mechanism besides the vapour-capture mechanism above. In conclusion, PM sorbent could effectively reduce the emission of fine PM from the combustion of coal. However, up to now, only few PM sorbents are available in public and they might hardly meet the various combustion conditions (e.g., combustion temperature, atmosphere, coal property, etc.). Therefore, more efforts are needed to seek new effective PM sorbents in developing the in-furnace PM control technique. As reviewed above, PM sorbents reduce the emission of ultrafine PM mainly via capturing the PM precursor (e.g., vapourized mineral matter such as alkali metal vapour, SOx, etc.) during the coal combustion. In this regard, sorbents/additives that are developed to capture gaseous mineral species at high temperature may also be used as PM sorbents. Such additives include clay minerals (mainly containing SiO2 and Al2O3) and functional materials (e.g., volcanic ash, coal fly ash, etc.) used in solving the ash deposition, corrosion and slagging problems involved in the combustion/gasification of coal or biomass [20–24]. For example, Shadman and Punjak et al. [25,26], Lee et al. [27] reported that emathlite, silica gel, alumina, diatomaceous earth, activated bauxite, attapulgus clay and dolomite could react with the vapourized mineral matter (e.g., NaCl, Pb, Cd, etc.) and fix them in the solid particles as aluminosilicates. And their performance appeared to be depended on the Si/Al content and their structures. Other materials such as bentonite, holloysite, bauxite and coal fly ash were also used to capture Na and K in biomass combustion and IGCC (Integrated Gasification Combined Cycle) [28–30]. Generally, above sorbent materials have distinctive physical properties (e.g., melting temperature, specific area, etc.) and/or chemical properties (e.g., reactivity, capacity, etc.), which facilitate their reaction with the vapourized mineral matter and are believed to bring them the ability to inhibit the partitioning of mineral matter into ultrafine PM. However, as far as we know, there are still no studies quantita-


tively evaluate the effects of above materials on the formation of ultrafine PM during coal combustion and their potential of being PM sorbent. In this paper, the effects of five clay minerals (montmorillonite, attapulgite, rectorite, vermiculite and wollastonite), one volcanic rock (pumice), one metal oxide (anatase) and the wellstudied sorbent-kaolinite on the emission of the ultrafine PM are studied. First, yields of ultrafine PM from the combustion of coal with each sorbent were compared with that of the blank run without sorbent addition to evaluate the performance of sorbents. And then, treatment of each sorbent with alkali metal vapour were carried out to ascertain the possible reactions and determine their Na capture capacities under the experimental conditions, which together with the particle size distribution, composition and morphology of PM help explore the reduction mechanism of the sorbents on ultrafine PM. Finally, properties led to the better performance of selected sorbents than the others were investigated. 2. Experimental 2.1. Materials A Chinese lignite coal (denoted as HLHC) with the particle size between 45 and 90 lm was used in the experiments. The ash content of coal HLHC is 21%, and other properties of coal are shown in Table 1. As can be seen, coal HLHC has a relatively high Na content. In total, 8 mineral materials were involved in the present study. They were 6 clay minerals (i.e., kaolin, montmorillonite, attapulgite, rectorite, vermiculite and wollastonite), a volcanic rock (pumice) and a metal oxide (anatase or titanium dioxide). The formula, Sauter mean diameter [D(3, 2)] and specific surface area (SSA) of each sorbent were listed in Table 2 and their elemental composition and minerals were listed in Table S1 in the supplementary material. Briefly, kaolin (KAO), montmorillonite (MON), vermiculite (VER) and rectorite (REC) were layer structure clay minerals with Si and Al as the primary components; meanwhile, they were of different SiO2/Al2O3 ratios. Anatase (ANA) was also of a layer structure while its primary component was titanium (Ti). Pumice (PUM) was a typical porous volcanic rock, which was mainly composed of Si and widely used in the water purification. Sorbent candidates used in the study were obtained from different mines, and few impurities were in them except for some SiO2 and FeS2 (see Fig. S1). Prior to the experiment, sorbents were dried (298 K), pulverized and screened, and the portion of a narrow size distribution was used. As shown in Table 2, Sauter mean diameters of the used sorbents were mostly located in the size range of 3.7–10.7 lm. The specific surface areas of the materials differed significantly from each other (1–140.99 m2/g). In the second part experiments, sodium acetate (NaAc, chemical pure) powders were used. 2.2. Combustion of coal/sorbent blends and PM sampling Two part experiments were carried out successively. In the first part, pulverized coal/sorbent blends as well as the coal or sorbent alone were burned in a high temperature drop tube furnace (DTF) system to evaluate the effects of the each sorbent on the PM emission. As sketched in Fig. 1, DTF system consisted of a SANKI piezo bowl vibratory feeder, a corundum tube reactor, a gas distributor and a furnace controller, and more detailed information has been described in our previous reports [31,32]. Before the experiment, each sorbent was blended with the pulverized coal mechanically at a mass ratio of 5 g-sorbent/100 g-coal. During the experiment, coal/sorbent blends were fed into the furnace at a constant rate


Y. Xu et al. / Fuel 193 (2017) 72–80

Table 1 Analysis of the coal HLHC and coal ash in the experiments. Proximate analysis, wt.%, air dry basis M 5.80

V 47.52

Ultimate analysis, wt.%, air dry basis A 21.53

FC 25.15

SiO2 64.58

P2O5 0.95

C 49.96

H 5.84

Cl2O 0.5

K2O 1.08

N 1.07

S 0.62

Oa 15.18

TiO2 0.5

Fe2O3 3.14

CuO 0.05

Ash analysisb, wt% Na2O 1.72 a b

MgO 1.5

Al2O3 16.6

SO3 5.52

CaO 3.95

By difference. Ashed 673 K in O2.

Table 2 Sauter mean diameter [D(3, 2)] and BET specific surface area of the sorbents.

a b c d e f




D(3, 2)b





Kaolin Montmorillonite

1.43 10.16

15.64 59.30

3.06 n.a.e

– –


Attapulgite Rectorite Vermiculite Pumice Wollastonite Anatase

Al2Si2O5(OH)4 (Na, Ca)0.33(Al, Mg)2 (Si4O10)(OH)2nH2O (Mg, Al)2Si4O10(OH)4(H2O) (Na, Ca)Al4(Si, Al)8O20(OH)42H2O (Mg, Fe3+, Al)3(Si, Al)4O10(OH)24H2O – CaSiO3 TiO2

3.93 10.03 3.77 10.74 4.58 15.49

140.99 – 18.20 2.38 1.00 15.28

n.a. 0.29 – n.a. 0.06 7.06

– – – – 1819 2116

From Mineralogical Society of America, http://www.handbookofmineralogy.org/index.html. Sauter mean diameter, lm. BET specific surface area, m2/g. BET specific surface area of the sorbents after heat treatment at 1500 °C, m2/g. The value is less than 0.01 m2/g. Melting temperature, K (Richet, 1998).

Fig. 1. Sketch of the drop tube (DTF) furnace system in combination of the LPI sampling system and the high temperature metal mesh sampling system. Inserted pictures are the spent sorbent KAO after the reaction with NaAc.

of 0.2 g/min (on the coal basis) by the feeder, and then entrained into the furnace by a gas stream of 5 L/min simulated air (21 vol. % O2 + 79 vol.% N2). The blends burned out in the furnace at the wall temperature of 1773 K. with a residence time of 1.2 s. Experiments of coal without sorbent addition were conducted as a blank run and four repeats were conducted under each condition. Standard deviations were calculated based on 4 independent parallel experiments and shown in the results as error bars. Combustion products were extracted from the bottom of the furnace and the PM in them was collected by a LPI sampling system. As shown in Fig. 1, the LPI sampling system contained a Cyclone, a Dekati low pressure impactor (LPI), a pressure gauge (LEO-2) and a vacuum pump (SV-25B) [8,32]. During PM sampling, PM with the aerodynamic diameter >10 lm (PM>10) were first separated in the Cyclone and then PM less than 10 lm (PM10) passed

into LPI downstream of it. In the LPI, PM10 was fractionated according to the particle size and collected on the aluminum foils coated with grease (Apiezon grease H) or polycarbonate membranes (Whatman co.). PM samples collected on the foils were weighed on a microbalance (Sartorius M2P). All of LPI, cyclone and pipes connecting them were externally heated (403 K) to avoid possible condensation of gaseous species [33]. Moreover, PM samples collected on the membrane were analysed with the X-ray fluorescence (XRF, EDAX Inc.) and field emission scanning electron microscope coupled with energy-dispersive X-ray spectrometry (FESEM-EDX, Sirion 200, FEI. co.) to determine the chemical composition of bulk sample as well as the composition and micromorphology of some selected particles in the sample. Prior to the observation under FESEM, samples were coated with a thin layer of Pt to improve the conductivity.

Y. Xu et al. / Fuel 193 (2017) 72–80

2.3. Reactions between sorbent candidates with alkali metal vapour and SO2 In the second part, blends of NaAc powder and each sorbent were injected into the same drop tube furnace used in the first part, and reacted in the simulated air (21 vol.% O2 + 79 vol.% N2) at 1500 °C to elucidate the reactions between sorbent and Nacontained vapour and the Na capture capacity of each sorbent at high temperature. What’s more, some selected sorbent candidates were also treated in the simulated air (21 vol.% O2 + 79 vol.% N2) with extra SO2 (2000 ppm) at 1500 °C in the drop tube furnace to determine their SO2 capture capacity. Before the experiment, NaAc powder was mixed mechanically with each sorbent at a mass ratio of 3.2 g-NaAc/100 g-sorbent. In the experiment, the NaAc/sorbent blends or sorbent alone were fed at a rate of 0.2 g/min and entrained into furnace (1773 K) by simulated air (8 L/min). Reaction products were collected via a self-designed hightemperature sampling probe inserted into the furnace. As sketched in Fig. 1, the sampling probe with a funnelled metal mesh set on its top, was inserted into the reactor tube from the furnace bottom. In this way, flue gas was sucked into the probe and the spent sorbent particles in it were collected on the mesh. The metal mesh was woven with 316 L stainless steel wires and the unused mesh has pores of 5.5 lm. Samples were collected at 1223 K to avoid the condensation of Na-contained vapour or SO2 on them, which helps focus on the capture of Na in the sorbents via the chemical fixation. The samples collected in the mesh at 1223 K may continue to react with the gaseous mineral species during the sampling, hence the capture capacity of the sorbents actually represented the Na captured in the sorbents during the whole 5 min sampling period. As shown in Fig. 1, no visible reactions between the mesh and the sorbents were observed in the experiments. The sorbents before and after the reaction were characterized with Fourier Transform Infrared Spectroscopy (FTIR) and X-ray Diffraction (XRD) to determine the reactions between each sorbent candidate and gaseous mineral species under the experimental condition. In FTIR determination, sample was mechanically mixed with potassium bromide (1.3 mg/130 mg-KBr) and pressed into a pellet. Furthermore, raw sorbents and their products were digested in the HNO3/HF (7 mL/2 mL) solution via a microwave digestion system. The contents of Na and S were determined by the Atomic Absorption Spectroscopy (AAS) (TAS-990) and the capture capacities of Na and S of the sorbents was calculated by difference [31]. Additionally, the spent sorbent ANA collected was put in water and heated at 60 °C for 12 h. The sludge was filtered and the content of soluble Na in the solution was determined via an ion chromatography(IC) to characterize the ratio of Na deposited to ANA particles physically. Results showed that little (2%) Na was deposited to ANA particles physically under the experimental and sampling conditions. The specific surface areas of the sorbents after heat treatment at 1500 °C in the DTF were also determined and the results are listed in Table 2.

3. Results and discussion 3.1. Effects of sorbents on the size distribution of particulate matter Fig. 2a and b presented the particle size distributions (PSDs) of the PM emitted from the combustion of coal alone and coal/sorbent blends. Additionally, the PSD of the sorbent KAO treated alone under the same condition was also shown. It can be seen that some PM was derived from the sorbent. However, as the derived PM was far less than that generated from the coal or coal/sorbent blends, this portion of PM was ignored in the discussion below.


PM generated from the combustion of coal alone and coal/sorbent blends were of three modal size distribution, consistent with the early reports [34,35]. As shown in Fig. 2, it consisted of an ultrafine modal PM below 0.2 lm (PM0.2), a mid-modal PM in 0.2–2.0 lm (PM0.2–2) and a coarse modal PM larger than 2.0 lm (PM>2). Compared with the blank run where coal was burned alone, sorbent addition into the furnace generally increased the emission of PM>2. However, the effects of the sorbent addition on the emission of PM2 (PM less than 2 lm) differed significantly from each other. Briefly, sorbent KAO, MON, ATT, WOL and ANA reduced the emission of PM2. Meanwhile, sorbent PUM, VER or REC had no obvious reduction effects on the PM2. Most importantly, the tested sorbents significantly changed the distribution of ultrafine PM (PM0.2). In this study, particular attention was paid on the effects of sorbents on the most concerned ultrafine PM. As shown in Fig. 2c and d, the peak size of the ultrafine mode remained at 0.08 lm with or without sorbent addition. However, the mass yields of the ultrafine PM varied significantly when sorbent was added. 3.2. Effects of sorbents on the yield of ultrafine PM and PM2.5 The effects of each sorbent on the formation of the PM were first discussed based on the mass yield and the reduction ratio. The PM0.2 and PM2.5 reduction ratios were calculated according Eq. (1), and the results are shown in Fig. 3 along with their yields.

Reduction ratio ¼ ðPMwithout sorbent  PMwith sorbent Þ=PMwithout sorbent  100 ð1Þ As can be seen in Fig. 3, 0.23 mg ultrafine PM was generated during the combustion of coal HLH under the experimental conditions. When sorbent KAO was added, the mass yield of the ultrafine PM reduced from 0.23 mg-PM0.2/g-coal to 0.17 mg-PM0.2/g-coal, corresponding to a reduction ratio of 22%. The reduction effect of sorbent KAO on ultrafine PM in the experiment was lower than that reported by Chen et al. [10], which was most possibly due to the lower addition ratio (23% vs. 240% ash-basis) and higher combustion temperature (1773 K vs. 1173–1573 K) in the present study. The results confirmed the findings of Chen et al. [10] and further implied that kaolin was not effective enough under hightemperature combustion conditions in real steam boilers. In the study, sorbents MON, ATT, WOL and ANA exhibited good performance in reducing the ultrafine PM, with PM0.2 reduction ratios of 26.27%, 24.27%, 22.05% and 39.42%, respectively. By contrast, sorbents PUM and VER exhibited weak reduction effects on the ultrafine PM (10.51% and 4.89% respectively), and the addition of sorbent REC even led to an increase of PM0.2 yield (7.93%). Most importantly, sorbent ANA showed the best removal performance on ultrafine PM with a PM0.2 reduction ratio of 39%. Compared with other sorbent candidates, sorbent ANA appeared to be a new effective ultrafine PM sorbent. At the same time, adding sorbent ANA also reduced the emission of PM2.5 from 1.79 mg/g-coal to 1.50 mg/g-coal, which corresponded to a reduction efficiency of 16.22%. The reduction efficiency of sorbent ANA on the PM2.5 was a little lower than that of sorbent KAO (21.47%) meanwhile higher than those of the other tested sorbents. The above results indicated that sorbent ANA was effective in reducing the emission of both ultrafine PM and PM2.5. In Section 3.3, effects of sorbent ANA on the formation of ultrafine PM were further analysed with the chemical composition. 3.3. Effects of sorbent ANA on the formation of ultrafine PM Chemical compositions of the ultrafine PM (e.g., PM0.1–0.2) produced from the combustion of coal HLH alone and coal/ANA blends


Y. Xu et al. / Fuel 193 (2017) 72–80

Fig. 2. Particle size distributions of (a and b) the PM10 and (c and d) the PM0.2 generated during coal HLHC combustion with the sorbents compared with the blank run without sorbent.

Fig. 4. Chemical composition of PM in 0.12–0.21 lm (PM0.1–0.2) and PM in 1.95– 3.08 lm (PM1.95–3.08) formed in the combustion of coal HLH with the addition of sorbent ANA compared with that from the combustion of coal HLH alone.

Fig. 3. Yields and reduction ratios of the (a) PM0.2 and (b) PM2.5 from the combustion of coal HLH alone and coal/sorbent blends under the experimental condition.

were shown in Fig. 4. Obviously, Na and S were the primary components of the ultrafine PM (e.g., PM0.1–0.2), which was consistent with the previous studies [3,17,36,37]. As confirmed in these studies, ultrafine PM was primarily formed through the vapourizationnucleation of mineral matter in coal during its combustion [38,39]. Volatile mineral matter such as Na vapourized at high temperature

and thereby enriched in the ultrafine PM. As for the coarse mode PM (e.g., PM1.95–3.08), Si and Al were the primary component and the addition of sorbent ANA led to the increase of Ti in it (see Fig. 4). As can be seen in Fig. 4, contents of Na and S in the ultrafine PM significantly changed when sorbent ANA was added during the combustion of coal. The results indicated that the change of the ultrafine PM yield was possibly resulted from the reduced partitioning of Na/S-contained gaseous species into the ultrafine PM by the sorbents, similar to what happened when sorbent KAO was added [9,10,17,18]. As claimed by Wendt et al. [17,18] and our previous studies [12], kaolin would react with the alkali metal vapour - the primary precursor of the ultrafine PM - and capture them into the sorbent particles, and thereby inhibited the ultrafine PM formation. To ascertain the possible reactions between sorbent

Y. Xu et al. / Fuel 193 (2017) 72–80

ANA and ultrafine PM precursor (mainly alkali metal vapour and SOx), sorbent ANA was further treated with NaAc and SO2 under the same experimental conditions. And the reaction products were characterized with FTIR and XRD. As with previous studies [12,19,40], sodium acetate was selected as a modal compound to simulate the partitioning of Na in coal during combustion. NaAc would burn at the high temperature and generate Na-contained vapour such as Na2O and NaOH [12,38,39]. Results in Fig. 5a showed that new peak at 973 cm1 occurred when sorbent ANA (anatase) was treated with NaAc at high temperature, which indicated that sorbent ANA reacted with Nacontained vapour and formed titanates under the experimental conditions. This result was further confirmed by the XRD patterns in Fig. 6, where sodium titanate (Na2Ti6O13, peaks denoted by S) appeared when sorbent ANA was reacted with NaAc. By contrast, no notable change was observed when sorbent ANA was treated in the SO2-contained atmosphere (see Figs. 5b and 6), indicating that sorbent ANA was not reacted with SO2 significantly under the experimental conditions. Above results suggested that the reduction effects of sorbent ANA on the formation of ultrafine PM were mainly resulted from the capture of alkali metal vapour (e.g., Na2O, NaOH) which inhibited their partitioning into the ultrafine PM. What’s more, as less alkali metals migrated into ultrafine PM, less sulphates/sulphites were formed, which further reduced the partitioning of S into the ultrafine PM and resulted in the decreased S content in the ultrafine PM derived from the combustion of coal/ANA blends (see Fig. 4). SiO2/Al2O3 ratio of PM0.12–0.21 also reduced when sorbent ANA was added, indicating less Si migrated into ultrafine PM. The change of Si content in ultrafine PM is supposed to be resulted from the reduced formation of sodium silicates as more Na was captured by sorbent ANA meanwhile further verifying experiments on this aspect were not conducted in the present study. And finally, the changed portioning behaviour of mineral matter resulted in the reduced formation of ultrafine PM (see Figs. 2 and 3).


Fig. 6. XRD patterns of sorbent ANA heated at 1273 K in muffle furnace for 1 h, products of sorbent ANA reacted with NaAc/SO2 at 1773 K in DTF and products of sorbent ANA reacted with NaAc at 1273 K in muffle furnace. In the figure, R is rutile (TiO2, No. 96-900-7532), S is a sodium titanate (Na2Ti6O13, No. 01-073-1398).

3.4. Comparison of ANA with other sorbents The capture ability of each sorbent for alkaline metal vapour was further characterized. And the correlations between the PM0.2 reduction ratio and the Na captured in each sorbent were shown in Fig. 7. It can be seen that the amount of Na fixed by the sorbent varied from 2.35 mg-Na/g-sorbent to 7.84 mg-Na/gsorbent under the experimental condition. It is noteworthy that the PM0.2 reduction ratio of the tested sorbents were positively related to the amount of Na captured in them, which further confirmed that the reaction and fixation of Na-containing vapour in the sorbents resulted in the reduction of ultrafine PM when the

Fig. 7. Correlations between the PM0.2 reduction ratio and the Na captured in the sorbent in the experiments.

sorbents were added. What’s more, it further suggested that the better PM removal performance of sorbent ANA, compared with other sorbents, was resulted from its higher Na capture ability. The Na capture abilities of the tested sorbents and their performance on the ultrafine PM reduction differed from each other sig-

Fig. 5. FT-IR patterns of (a) sorbent ANA heated at 1273 K in simulated air in muffle furnace for 1 h, products of sorbent ANA reacted with NaAc at 1773 K in simulated air in DTF and products of sorbent ANA reacted with NaAc at 1273 K in muffle furnace; (b) products of sorbent ANA reacted with SO2 at 1273 K in muffle furnace.


Y. Xu et al. / Fuel 193 (2017) 72–80

way, the added sorbent ANA particles (see the irregular particles in Fig. 9) can capture the gaseous PM formation precursor, and inhibit their migration into ultrafine PM.

Fig. 8. Composition of sorbents KAO, REC and ANA, and composition of PM0.12–0.21 generated from the combustion of coal HLH alone and that from the combustion of coal blends coal/KAO, coal/REC and coal/ANA.

nificantly. And the disparities were supposed to be related to their distinctive behaviour at the high experimental temperature as a result of their different composition and the microstructure. First, sorbents MON/ATT/REC/PUM/VER and KAO are clay minerals, and the main composition of which are Si and Al. The reactions between Na2O vapour and clay minerals are similar to those of sorbent KAO, which can be expressed in Eqs. (2) and (3). As shown in Table S1, sorbents MON and ATT had a higher content of Si, which can facilitate the capture of Na2O vapour through Eq. (3). As seen in Fig. 8, the contents of Si and Al of sorbent REC are similar to those of KAO, however, there is a much higher content of S. The sulphur released at the high temperature and produced new ultrafine PM enriched in S, and that was supposed to be the reason why the addition of sorbent REC caused the increase of PM0.2 yield. Different from other sorbent candidates, anatase (TiO2) is the main composition of sorbent ANA and TiO2 would react with Na2O at elevated temperatures via reaction (4) [41,42]. In this

Na2 O þ SiO2  nAl2 O3 ¼ Na2 O  SiO2  nAl2 O3


Na2 O þ SiO2 ¼ Na2 O  SiO2 þ H2 O


Na2 O þ nTiO2 ¼ Na2 O  nTiO2 þ H2 O


Second, the structure features and physical properties of the sorbents also played an important role in capturing ultrafine PM precursors at high temperature. Similar to kaoline, anatase was also of layered structure. The layered structure would facilitate the diffusion of reactant species in the sorbent particle, which contributed to the higher reaction rate and capture capacity. The BET specific surface areas of the raw sorbents before the experiment were determined and the results were listed in Table 2 in Section 2.1. As can be seen that the SSA of sorbent ANA was 15.28 m2/g, which was in the range of SSAs of all the tested sorbents (1.00–140.99 m2/g). However, no obvious relationship was observed between the ultrafine PM reduction performance of the sorbents and the specific surface area of the raw sorbents. The BET specific surface areas of the sorbents reduced after heat treatment and sorbent ANA had the largest specific area, which implied that the performance of the sorbents was more related to their behaviours at high combustion temperature. Fig. 9 exhibited the micromorphology of PM in the size range of 1.95–3.08 lm, collected in the experiments when coal/ANA and coal/KAO blends were burned. Some particles of sorbent ANA were observed in the PM samples (see Fig. 9a). And it showed that sorbent ANA particles were not sintered and remained irregular shapes after undergoing the high temperature, which provided a larger surface to contact and capture those gaseous ultrafine PM precursor during the coal combustion. In comparison, sorbent KAO particles were mostly molten (see Fig. 9b), which was bad to the gas-to-solid reaction above. In conclusion, the high sintering temperature of sorbent ANA in the combustion condition was also an essential factor that led to its good performance on the reduction of PM0.2. The

Fig. 9. Microscope and composition of PM with aerodynamic diameters in 1.95–3.08 lm generated from the combustion of coal HLH with addition of sorbent ANA and KAO.

Y. Xu et al. / Fuel 193 (2017) 72–80

Ti-based sorbent with a good performance in capturing ultrafine PM and Na-contained vapour under the high temperature combustion conditions indicated its potential of being a high temperature sorbent. 4. Conclusions The present work aims to search for new effective sorbents which can reduce the ultrafine PM emission during the coal combustion. At first, screening experiments were carried out to determine the effects of each sorbent on the emission and composition of ultrafine PM. And then further combustion of NaAc and sorbents was conducted to explore the reduction mechanism of the sorbents. The main conclusions are: (1) A novel Ti-based PM sorbent was screened out of 8 mineral candidates which has a better performance in reducing the ultrafine PM formation than the other sorbents (kaolin, montmorillonite, attapulgite, rectorite, vermiculite, pumice and wollastonite). (2) The Ti-based sorbent (anatase) had an ultrafine PM removal efficiency of 39% under the experimental conditions, which was higher than that of the well-studied kaolin (22%). (3) The Ti-based sorbent reduced the PM0.2 emission mainly through capturing Na-contained vapour. Its distinctive layered structure and high sintering temperature resulted in the higher Na fixation ability and, thus, better performance in reducing ultrafine PM. It should be noted that there are still many challenges in using this PM sorbent and more work is stilled needed. On one hand, the impacts of sorbent addition on the operation of boiler should be characterized such as the coal burnout, the ash deposition and slagging, the erosion/corrosion of the heat transfer surface, etc. On the other hand, adding PM sorbent will increase the yield of bulk ash and/or change its chemical and physical properties (e.g., chemical composition, electrical resistivity, etc.), which may affect the operation of the equipped flue gas cleaning devices. To minimize the possible adverse effects, the suitable adding amount and adding strategy should be found out. Acknowledgements The authors acknowledge the financial support of the National Natural Science Foundation of China (51476064, 51661125011) and the National Basic Research Program of China (2013CB228501). The authors also thank the Fundamental Research Funds for the Central Universities (HUST: No. CX15018) and the support of the Analytical and Testing Center at the Huazhong University of Science and Technology. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2016.12.043. References [1] Xu M, Yu D, Yao H, Liu X, Qiao Y. Coal combustion-generated aerosols: formation and properties. Proc Combust Inst 2011;33(1):1681–97. [2] Huang R, Zhang Y, Bozzetti C, Ho K, Cao J, Han Y, et al. High secondary aerosol contribution to particulate pollution during haze events in China. Nature 2014;514(7521):218–22. [3] Linak WP, Yoo JI, Wasson SJ, Zhu W, Wendt J, Huggins FE, et al. Ultrafine ash aerosols from coal combustion: characterization and health effects. Proc Combust Inst 2007;31(Part 2):1929–37.


[4] Yao Q, Li SQ, Xu HW, Zhuo JK, Song Q. Reprint of: Studies on formation and control of combustion particulate matter in China: a review. Energy 2010;35 (11):4480–93. [5] Wang C, Liu X, Li D, Si J, Zhao B, Xu M. Measurement of particulate matter and trace elements from a coal-fired power plant with electrostatic precipitators equipped the low temperature economizer. Proc Combust Inst 2015;35 (3):2793–800. [6] Liu X, Xu Y, Fan B, Lv C, Xu M, Pan S, et al. Field measurements on the emission and removal of PM2.5 from coal-fired power stations: 2. studies on two 135 MW CFB boilers respectively equipped with electrostatic precipitator and hybrid electrostatic filter precipitator. Energy Fuel 2016;30(7):5922–9. [7] Xu Y, Liu X, Zhang Y, Sun W, Zhou Z, Xu M, et al. Field measurements on the emission and removal of PM2.5 from coal-fired power stations: 3. direct comparison on the PM removal efficiency of electrostatic precipitators and fabric filters. Energy Fuel 2016;30(7):5930–6. [8] Lu Q, Yang Z, Zheng C, Li X, Zhao C, Xu X, et al. Numerical simulation on the fine particle charging and transport behaviors in a wire-plate electrostatic precipitator[J]. Adv Powder Technol 2016;27(5):1905–11. [9] Chen J, Yao H, Zhang PA, Xiao L, Luo G, Xu M. Control of PM1 by kaolin or limestone during O2/CO2 pulverized coal combustion. Proc Combust Inst 2011;33(2):2837–43. [10] Qu C, Xu B, Wu J, Liu J, Wang X. Effect of CaO additive on formation of PM2.5 under O2/CO2 atmosphere during coal combustion. Second international conference on mechanic automation and control engineering (MACE); Hohhot; 2011. p. 2112–5. [11] Si J, Liu X, Xu M, Sheng L, Zhou Z, Wang C, et al. Effect of kaolin additive on PM2.5 reduction during pulverized coal combustion: importance of sodium and its occurrence in coal. Appl Energy 2014;114:434–44. [12] Wei Y, Wang Q, Zhang L, Awaya I, Ji M, Li H, et al. Effect of magnesium additives on PM2.5 reduction during pulverized coal combustion. Fuel Process Technol 2013;105:188–94. [13] Ninomiya Y, Wang Q, Xu S, Mizuno K, Awaya I. Effect of additives on the reduction of PM2.5 emissions during pulverized coal combustion. Energy Fuels 2009;23(7):3412–7. [14] Zhou Z, Liu X, Zhao B, Shao H, Xu Y, Xu M. Elemental mercury oxidation over manganese-based perovskite-type catalyst at low temperature[J]. Chem Eng J 2016;288:701–10. [15] Wendt JOL, Lee SJ. High-temperature sorbents for Hg, Cd, Pb, and other trace metals: mechanisms and applications. Fuel 2010;89(4):894–903. [16] Gale TK, Wendt JOL. In-furnace capture of cadmium and other semi-volatile metals by sorbents. Proc Combust Inst 2005;30(2):2999–3007. [17] Gale TK, Wendt JOL. Mechanisms and models describing sodium and lead scavenging by a kaolinite aerosol at high temperatures. Aerosol Sci Technol 2003;37(11):865–76. [18] Gale TK, Wendt JOL. High-temperature interactions between multiple-metals and kaolinite. Combust Flame 2002;131(3):299–307. [19] Uberoi M. High-temperature removal of metal vapors by solid sorbents [dissertation]. Tucson: University of Ariozona; 1990. [20] Punjak WA, Uberoi M, Shadman F. Control of ash deposition through the high temperature adsorption of alkali vapors on solid sorbents. ACS division of fuel chemistry, preprints of papers presented at the ACS national meeting in dallas, Texas, 1989. [21] Ho TC, Chen J, Shukla S, Hopper JR. Metal capture during fluidized bed incineration of solid wastes. J Hazard Mater 1989;22(2):257–8. [22] Punjak WA, Shadman F. Aluminosilicate sorbents for control of alkali vapors during coal combustion and gasification. Energy Fuels 1988;2(5):702–8. [23] Wang L, Hustad JE, Skreiberg O, Skjevrak G, Gronli M. A critical review on additives to reduce ash related operation problems in biomass combustion applications. Energy Procedia 2012;20:20–9. [24] Uberoi M, Shadman F. High-temperature removal of cadmium compounds using solid sorbents. Environ Sci Technol 1991;25(7):1285–9. [25] Uberoi M, Punjak WA, Shadman F. The kinetics and mechanism of alkali removal from flue gases by solid sorbents. Prog Energy Combust Sci 1990;16 (4):205–11. [26] Lee SHD, Johnson I. Removal of gaseous alkali metal compounds from hot flue gas by particulate sorbents. J Eng Power 1980;102(2):397–402. [27] Hardy T, Kordylewski W, Moscicki K. Aluminosilicate sorbents for control of KCl vapors in biomass combustion gases. J Power Technol 2013;93(1):37–43. [28] Bläsing M, Müller M. Investigation of the effect of alkali metal sorbents on the release and capture of trace elements during combustion of straw. Combust Flame 2013;160(12):3015–20. [29] Waindich A, Muller M. Alkali removal at 1400 °C under gasification conditions. Fuel 2013;116:889–93. [30] Wang C, Liu X, Li D, Wu W, Xu Y, Si J, et al. Effect of H2O and SO2 on the distribution characteristics of trace elements in particulate matter at high temperature under oxy-fuel combustion. Int J Greenhouse Gas Control 2014;23:51–60. [31] Xu Y, Liu X, Zhou Z, Sheng L, Wang C, Xu M. The role of steam in silica vaporization and ultrafine particulate matter formation during wet oxy-coal combustion. Appl Energy 2014;133:144–51. [32] Gao X, Wu H. Effect of sampling temperature on the properties of inorganic particulate matter collected from biomass combustion in a drop-tube furnace. Energy Fuel 2010;24(8):4571–80. [33] Xu Y, Liu X, Zhang P, Guo J, Han J, Zhou Z, Xu M. Role of chlorine in ultrafine particulate matter formation during the combustion of a blend of high-Cl coal and low-Cl coal[J]. Fuel 2016;184:185–91.


Y. Xu et al. / Fuel 193 (2017) 72–80

[34] Yu DX, Xu MH, Yao H, Liu XW, Zhou K. A new method for identifying the modes of particulate matter from pulverized coal combustion. Powder Technol 2008;183(1):105–14. [35] Gao X, Rahim MU, Chen X, Wu H. Significant contribution of organically-bound Mg, Ca, and Fe to inorganic PM10 emission during the combustion of pulverized Victorian brown coal. Fuel 2014;117:825–32. [36] Li G, Li S, Huang Q, Yao Q. Fine particulate formation and ash deposition during pulverized coal combustion of high-sodium lignite in a down-fired furnace. Fuel 2015;143:430–7. [37] Quann RJ, Sarofim AF. Vaporization of refractory oxides during pulverized coal combustion. Symposium (international) on combustion, vol. 19. Elsevier; 1982. p. 1429–40 (1). [38] Neville M, Quann RJ, Haynes BS, Sarofim AF. Vaporization and condensation of mineral matter during pulverized coal combustion. Symposium (international) on combustion, vol. 18. Elsevier; 1981. p. 1267–74 (1).

[39] Mwabe PO, Wendt JOL. Mechanisms governing trace sodium capture by kaolinite in a downflow combustor. Symposium (international) on combustion, vol. 26. Elsevier; 1996. p. 2447–53 (2). [40] Sauvet A, Baliteau S, Lopez C, Fabry P. Synthesis and characterization of sodium titanates Na2Ti3O7 and Na2Ti6O13. J Solid State Chem 2004;177 (12):4508–15. }rösi L, Papp S, Csapó E, Meynen V, Cool P, Dékány I. A short solid-state [41] Ko synthesis leading to titanate compounds with porous structure and nanosheet morphology. Microporous Mesoporous Mater 2012;147(1):53–8. [42] Papp S, Kõrösi L, Meynen V, Cool P, Vansant EF, Dékány I. The influence of temperature on the structural behaviour of sodium tri-and hexa-titanates and their protonated forms. J Solid State Chem 2005;178(5):1614–9.