J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

ScienceDirectwww.e l sev i e r . com/ loca te / j es

Hierarchical Ag-SiO2@Fe3O4 magnetic compositesfor elemental mercury removal from non-ferrousmetal smelting flue gas

Yongpeng Ma1, Bailong Mu1, Xiaojing Zhang1, Hongzhong Zhang1, Haomiao Xu2,⁎,Zan Qu2, Li Gao3

1. Henan Collaborative Innovation Center of Environmental Pollution Control and Ecological Restoration, School of Material and ChemicalEngineering, Zhengzhou University of Light Industry, No. 136, Science Avenue, Zhengzhou 450001, China2. School of Environmental Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China3. College of Resources and Environmental Science, Ningxia University, Yinchuan 750021, China

A R T I C L E I N F O

⁎ Corresponding author. E-mail: xuhaomiao@s

https://doi.org/10.1016/j.jes.2018.11.0141001-0742 © 2018 The Research Center for Ec

A B S T R A C T

Article history:Received 14 May 2018Revised 18 October 2018Accepted 9 November 2018Available online 1 December 2018

Hierarchical Ag-SiO2@Fe3O4 magnetic composites were selected for elemental mercury(Hg0) removal from non-ferrous metal smelting flue gas in this study. Results showed thatthe hierarchical Ag-SiO2@Fe3O4 magnetic composites had favorable Hg0 removal ability atlow temperature. Moreover, the adsorption capacity of hierarchical magnetic composite ismuch larger than that of pure Fe3O4 and SiO2@Fe3O4. The Hg0 removal efficiency reachedthe highest value as approximately 92% under the reaction temperature of 150°C, while theremoval efficiency sharply reduced in the absence of O2. The characterization resultsindicated that Ag nanoparticles grew on the surface of SiO2@Fe3O4 support. The largesurface area of SiO2 supplied efficient reaction room for Hg and Ag atoms. Ag–Hg amalgamis generated on the surface of the composites. In addition, this magnetic material could beeasily separated from fly ashes when adopted for treating real flue gas, and the spentmaterials could be regenerated using a simple thermal-desorption method.© 2018 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences.

Published by Elsevier B.V.

Keywords:Elemental mercurySilver nanoparticlesMagnetic compositesNon-ferrous metalFlue gas

Introduction

Mercury (Hg), a hazardous and global environmental contam-inant with high volatility, bioaccumulation and persistence inthe environment, has drawn increasing attention in world-wide in recent years (Liu et al., 2017a, 2017c; Zhang et al.,2015;). It was reported that approximately 2320 tons ofmercury was emitted annually from anthropogenic activities(Ma et al., 2015). Moreover, China was regarded as the largestcontributor to global anthropogenic atmospheric mercuryemission. It was calculated that that China accounted 27% of

jtu.edu.cn (Haomiao Xu).

o-Environmental Science

the total emission (Pacyna et al., 2010; Pirrone et al., 2010).Among all the emission sources in China, non-ferrous metalsmelting industries contributed to approximately 40% (Liuet al., 2017b; Zhang et al., 2015). Especially, the lead, zinc andcopper smelting accounted for 86% of the mercury emissionsfrom non-ferrous smelting processes (Zhang et al., 2012).Global atmospheric mercury emissions from non-ferrousmetal smelters reached 310 tons in 2007, among which morethan 65% were emitted from China (Streets et al., 2005; Wu etal., 2012). The Minamata Convention on Mercury, a newworldwide treaty aimed to reduce anthropogenic mercury

s, Chinese Academy of Sciences. Published by Elsevier B.V.

112 J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

emission to environment and protect human health, wassigned on October 10, 2013, and came into force from August6, 2017 (Selin, 2014a, 2014b). Hence, it is urgent to controlmercury emission from non-ferrous metal smeltingindustries.

Compared with the coal-fired flue gas, the mercuryconcentration was high and fluctuate widely during theproduction process in non-ferrous metal smelting flue gas,which also contained high concentration of SO2 (Wang et al.,2010b). However, most of the previous studies focused on themercury removal from coal-fired flue gas but paid littleattention to the non-ferrous metal smelting flue gas (Louie,2005; Ma et al., 2015). Traditional mercury purificationtechnologies in non-ferrous smelters contained wet adsorp-tion and adsorptionmethods (Wang et al., 2010a). Consideringthe high concentration of Hg0 in non-ferrous smelting fluegas, themost feasible method is the adsorption process whichenables the capture and reclamation of Hg0.

The adsorption methods highly dependent on the efficientadsorbents such as carbon-based materials (Liu et al., 2017b),metal oxides (Xiao et al., 2008), functionalized mesoporousmaterials (Cao et al., 2017), magnetic adsorbents and noblemetal adsorbents (Dong et al., 2017). However, most of theseadsorbents are limited in large-scale application for non-ferrous smelting flue gas due to their low capacity, unrecy-clable performance and high cost (Liu et al., 1998; Yang et al.,2011a). Usually, carbon-based materials take activated carbon(AC) as an example, can convert gaseous elemental mercury(Hg0) and oxidized mercury (Hg2+) to particle bound mercury(HgP). Then HgP can be removed by particulate control devices(e.g., Electrostatics precipitator or Fabric filter). However, thecost of AC is high, which limited its wide utilization. Sometransition metal oxides were recognized as potential adsor-bents (Yang et al., 2011c). Mn-based materials were indicatedto have large Hg0 adsorption capacities. The high valence ofMn in Mn-based oxides result in a prominent catalyticoxidation of Hg0 to Hg2+ over Mn-based oxides (Xu et al.,2015b). However, Mn-based oxides are easily poisoned todeactivity by SO2 in the flue gas.

Functionalized mesoporous materials also have excellentadsorption performances (Tomer et al., 2014). Mesoporoussilica was applied to pollution control in a previous study (Caoet al., 2017). The large specific surface area of mesoporoussilica provides sufficient space for reactions (Cao et al., 2017).Another advantage of using silica as support is its diffusionability, which avoids the agglomeration of the supportedmetal nanoparticles (Zhang et al., 2011). In addition, silica is anon-toxic material with the cross-linking structure (Du et al.,2006). However, most of adsorbents used for Hg0 removal fromflue gas mix into fly ash could not be separated, resulted inadsorbent loss and mercury secondary pollution (Huang et al.,2008). To resolve these problems, magnetically separableadsorbents are proposed in recent years because of theirsuperparamagnetic, strong adsorption ability and magneticsegregation properties (Du et al., 2017; Liang et al., 2017).

Noble metal-based adsorbents such as platinum (Pt), gold(Au) and silver (Ag)-based adsorbents, were indicated to haveHg0 removal performances with the amalgamation mecha-nism. Silver-based sorbents, such as Ag-graphene (Xu et al.,2015a), Ag-MagZ (Dong et al., 2017), were proved to have

excellent Hg0 adsorption efficiencies. When silver-basedsorbents used for mercury removal, it exhibited a possibilityfor Hg0 removal and Ag–Hg amalgams were generated. Then,mercury from the amalgamation could be recycled throughthermal treatment at a high temperature. In this paper, ahierarchical Ag-SiO2@Fe3O4 magnetic composite was synthe-sized and employed for Hg0 removal from simulated non-ferrous smelting flue gas. SiO2 was designated as the coatingof Fe3O4 to prepare SiO2@Fe3O4 composites, which formed aprotective film against instability of it. Also, themechanism ofthe interaction between Hg0 and Ag nanoparticles wasdiscussed.

1. Experimental section

1.1. Materials

FeCl3·6H2O (99.0%, AR grade, Sigma–Aldrich Co., Ltd.), ethyl-ene glycol (AR grade, Sigma–Aldrich Co., Ltd.), PEG 20000(Chemical pure grade, Sigma–Aldrich Co., Ltd.), anhydroussodium acetate (99.0%, AR grade, Kermel), ethyl alcohol(99.7%, AR grade, Sigma–Aldrich Co., Ltd.), tetraethylorthosilicate (Reagent grade, 98.0%, Aladdin), AgNO3 (ARgrade, 99.8%, Aladdin), polyvinyl pyrrolidone (AR grade,Aladdin) were used directly without further purification. TheSO2 (5%, Dalian Date Gas Co., Ltd.) and O2 (99.9%, Dalian DateGas Co., Ltd.) were stored in cylinders.

1.2. Preparation of Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4

Fe3O4 was synthesized using a typical method and SiO2 wascoated on the surface of Fe3O4. Ag nanoparticles were loadedon the surface of SiO2@Fe3O4 microspheres using an ionexchange method. The detailed synthesis process was con-ducted as follows: (1) Synthesis of monodisperse Fe3O4

microspheres. 1.35 g of FeCl3·6H2O was absolutely dissolvedin 40 mL of ethylene glycol, then 1.0 g of PEG 20000 was addedinto the above solution. Afterwards, 2.2 g of anhydroussodium acetate was added and the solution was kept stirringfor several minutes. When the mixture was totally dissolved,the solution was poured into a reactor with a Teflon innercontainer. The products were collected after heating at 200°Cfor 8 hr, and then washed with ethanol for several times,finally dried at 60°C for 6 hr in a vacuum oven. (2) Synthesis ofmagnetic Fe3O4@SiO2 microspheres. Typically, 0.2 g of above-prepared Fe3O4 microspheres was ultrasonic dispersed in amixture of 20 mL of ethyl alcohol and 4 mL of deionized waterto form a homogeneous solution. Then 0.8 mL of tetraethylorthosilicate was added into the solution under continuousmechanical stirring, followed by adding 1 mL of ammoniumhydroxide. After 3 hr of continuous mechanical stirring at25°C, magnetic separation was carried out for the Fe3O4@SiO2

products and then dried at 60°C for 3 hr. (3) Synthesis ofmonodisperse Ag-coated Fe3O4@SiO2 composite micro-spheres. 0.173 g of AgNO3 was dispersed in deionized waterto form a clear solution. Then, 0.173 g of polyvinyl pyrrolidone(PVP) was added into the homogeneous solution and ultra-sonically treated for 10 min. As a result, the mass ratio ofAgNO3:PVP was 1:1. Then, 1.0 g of magnetic Fe3O4@SiO2

113J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

microspheres synthesized in process (2) was added to thehomogeneous solution and ultrasonically treated for 10 min.After that, the products were transferred into a conical flaskequipped with the rotary evaporators. The solution wasconcentrated at a temperature of 50°C until it got thicken.Then, the above products were dried at 100°C for 5 hr until thesurface water was completely evaporated. Finally, the mate-rials were transferred into a muffle furnace for calcining 2 hrat 350°C.

1.3. Characterization of Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4

The microstructures of the materials were characterized byfield emission scanning electron microscopy (FESEM, JSM-7001F) and transmission electronmicroscopy (TEM, JEM-2100).And the micrographs were obtained in the bright-fieldimaging mode at an acceleration voltage of 200 kV. PowderX-ray diffraction (PXRD) was obtained to detect the crystalstructure of the prepared material, and the result wasrecorded on an X-ray diffractometer (DS, Advanced, Germany)with Cu Kα radiation. The scanning scale was in the 2θ rangefrom 10° to 80° with a scanning velocity of 10°/min. Thesurface element analysis of the material was performed by X-ray photoelectron spectroscopy (XPS), and the results wererecorded by ESCALAB 250 Xi XPS instrument (Thermo FisherScientific, USA) with the excitation source of Al Kα. Beforeeach processing of data, the C 1s line at 284.6 eV wascalibrated for the binding energy of the detected element.The hysteresis loops of the hierarchical magnetic particleswere measured by a Quantum Design MPMS-XL-7 magnetom-eter at magnetic field intensity from −15,000 to 15,000 Oe. Inaddition, the surface functional groups were characterized byFT-IR spectroscopy, and the 633 nm line of Ar+ laser was usedfor the excitation source.

1.4. Hg0 removal experiments

The bench-scale experimental equipments for Hg0 removalperformance consists of four parts: a gas distributingsystem, a reaction system of tubular furnace, an onlinedetection system and a tail gas treatment system, whichwas depicted in our previous study (Ma et al., 2017). Atotal gas flow of 350 mL/min was allowed for the exper-imental facility, which was conducted via mass flowcontrollers. Moreover, O2, SO2 and H2O vapor utilized inthis study were carried by pure N2. In addition, the Hg0

vapor was transported through a U-shaped glass tube,which was immersed in a thermostat water bath. Theinlet concentration of Hg0 in the testing gases wasapproximately 100 μg/m3, which was controlled byadjusting the temperature of the water bath and the flowrate of N2. The major experiments were carried out by atubular furnace with a temperature controller, which keptthe system at a precise temperature. The quartz tube withthe diameter of 4 mm was immobilized in the tubularfurnace where the simulated flue gas could pass through.The online detection system, which could analyze the inletand outlet Hg0 concentration, was monitored by a coldvapor atomic absorption spectrometer (CVAAS) mercurydetector and it was standardized by Lumex RA-915M

mercury analyzer to make sure its accuracy. The signalwas collected and recorded by a data transition andacquisition device (N2000, Zhida Ltd., China). A bottlecontained 5% KMnO4 solution and another bottle filledwith active carbon were used for the off-gas cleaning inthe tail-end equipment.

During each experiment, the mercury inlet gas bypassedthe adsorbent bed and passed into the side road until thecurve of inlet mercury concentration approached to flat.20 mg of As-prepared adsorbents were added into the quartztube with amounts of silica wool for immobilizing at a certainposition. The temperature of the reaction was set by atemperature controller in the tubular furnace. To investigatethe effect of temperature on the adsorption performance, thetemperature during the experiment was varied from 50 to200°C. Moreover, the reaction time of 120 min was permittedfor this experiment. And pure N2, 5%, 10% of O2 and 100, 200,500 ppm of SO2 were chosen to investigate the effects ofvarious gas components. The Hg0 removal efficiency wascalculated as Eq. (1):

Removal efficiency ¼ C0−CC0

� 100% ð1Þ

where the C0 is the inlet concentration of Hg0, and C is theoutlet concentration of Hg0.

In addition, the mercury temperature programmed de-sorption (Hg-TPD) experiment was also conducted after theHg0 removal experiment under the condition of simulated fluegas. In advance of each test, the adsorbent was treated via themixture of N2 + O2 accompanied with mercury vapor for30 min at 150°C to make it had a certain Hg0 concentration.Only pure N2 was the carrier gas for the Hg-TPD experiment.The temperature of tubular furnace in this system was setfrom 150 to 500°C with a speed of 5°C/min. The Hg-TPD couldbe accelerated at higher temperature, but the morphology ofthe Ag-SiO2@Fe3O4 might also be negatively impacted. Duringthe test process, the flow rate was kept at 350 mL/min, andthe mercury concentration was calculated by a CVAASmercury detector.

2. Results and discussion

2.1. Characterization of As-prepared materials

2.1.1. XRDPowder XRD was employed to measure the crystal struc-tures of As-prepared materials. The results are shown inFig. 1. As shown in Fig. 1 (a), all the diffraction peaks werein accordance with the standard Fe3O4 reflections (JCPDScard No. 75-1609), where these diffraction peaks are Fe3O4

(111), Fe3O4 (220), Fe3O4 (311), Fe3O4 (400), Fe3O4 (511) andFe3O4 (440), respectively. As shown in Fig. 1 (b), the broadband ranging from 2θ = 10° to 32° was the characteristicpeak of SBA-15 amorphous silica matrix (Cao et al., 2017).As silver coated on the surface of SiO2@Fe3O4, in thepattern of Ag-SiO2@Fe3O4 composites, another three dif-fraction peaks were marked to Ag (111), Ag (200) and Ag(220), comparing to the standard silver diffraction peaks(JCPDS No. 4-0783).

10 20 30 40 50 60 70 80

(c) Ag-SiO2@Fe3O4

(b) SiO2@Fe3O4

(a) Fe3O4

(511

)

(111

)

(a)

(111

)

(200

)

(220

)

(220

)

(440

)

(400

)

Inte

nsity

(a.u

.)

2 Theta (degree)

(311

)

(c)

(b)

Fig. 1 – X-ray diffraction patterns of the sorbents: (a) Fe3O4, (b)SiO2@Fe3O4 and (c) Ag-SiO2@Fe3O4.

114 J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

2.1.2. TEMThe nanostructures of Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4

were exhibited via TEM images, and the results are shown inFig. 2. Fe3O4 images were displayed in Fig. 2 (a) and (b), fromwhich a spherical morphology with a diameter of 200 ± 10 nmcould be clearly seen. Furthermore, the agglomeration phe-nomenon was not observed in the Fe3O4 micro-particles, andthe distance between the interplanar spacing was approxi-mately 0.245 nm, corresponding to the (111) crystal phase ofFe3O4. As shown in Fig. 2 (c) and (d), the SiO2 shell wassuccessfully coated on the surface of Fe3O4 micro-particles.Moreover, the thickness of SiO2 shell was about 20 ± 5 nm,which indicated the successful synthesis of SiO2@Fe3O4 core-shell structure. The Ag-SiO2@Fe3O4 images were exhibited inFig. 2 (e), (f) and (g). The uniform spherical morphologies ofAg-SiO2@Fe3O4 were exhibited with a diameter of about 260 ±10 nm. Therefore, hierarchical Ag-SiO2@Fe3O4 magnetic

0.245nm

20±5n

m

a c

b d100 nm 50 nm

20 nm2 nm

Fig. 2 – TEM images of (a) and (b) Fe3O4, (c) and (d)

composites were successfully synthesized based on TEM andXRD results.

The SEM patterns were used to investigate the morphol-ogies of Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4. It could beseen from Fig. 3 (a) and (b) that the Fe3O4 particles wereuniform in size. Fig. 3 (c) showed the image of SiO2@Fe3O4

synthesized by hydrolysis and condensation of tetraethylorthosilicate. As shown in Fig. 3 (d), the magnetic Ag-SiO2@Fe3O4 composites were obtained, and the Ag nanoparticleswere loaded on the surface of SiO2@Fe3O4 support. Thehierarchical Ag-SiO2@Fe3O4 magnetic composites were uni-formly in size, which was consistent with the TEM results.

2.1.3. FT-IRThe surface functional groups of the adsorbents were alsodetected by FT-IR spectra and the results are shown in Fig. 4.As shown in the spectra, the peaks centered at 3425 and1634 cm−1, reflecting the stretching and bending vibrationpeaks of hydroxyl group appeared, respectively (Xu et al.,2015b). The Fe–O characteristic absorption peak at centered at586.4 cm−1. The peaks at 1091.1, 950 and 806.7 cm−1 were thesymmetric stretching and bending absorption peaks of O–Si–O, revealing that the presence of element Si in the SiO2@Fe3O4

and Ag-SiO2@Fe3O4 nano-composites (Xu et al., 2015b).Moreover, Ag–O peak was not detected in the FT-IR spectraof Ag-SiO2@Fe3O4 magnetic particles. Therefore, it was spec-ulated that AgNO3 was converted to elemental Ag on thesurface of the particles.

2.1.4. MagnetizationTo investigate the magnetic characteristic of the preparedadsorbents, the hysteresis loops of the adsorbents weredetected via vibrating sample magnetometer (VSM). Theresults are shown in Fig. 5. The adsorbents exhibited softmagnetic characterization, and the saturation magnetizationof Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4 were 70.6, 30.4 and28.5 emu /g−, respectively. The reduction of saturation mag-netization of SiO2@Fe3O4 and Ag-SiO2@Fe3O4 was mainly due

e

gf200 nm

10 nm50 nm

SiO2@Fe3O4 and (e), (f) and (g) Ag-SiO2@Fe3O4.

Fig. 3 – SEM patterns of (a) and (b) Fe3O4, (c) SiO2@Fe3O4 and (d) Ag-SiO2@Fe3O4.

4000 3500 3000 2500 2000 1500 1000 500

Wavenumber (cm-1)

Fe3O4

SiO2@Fe3O4

Ag-SiO2@Fe3O4

3425

cm-1 16

34cm

-1

1091

.1cm

-1

806.

7cm

-1

586.

4cm

-1

950c

m-1

Fig. 4 – FT-IR spectra of Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4.

-15000 -10000 -5000 0 5000 10000 15000-80

-60

-40

-20

0

20

40

60

80

M (e

mu

g-1 )

H (Oe)

Fe3O4

SiO2@Fe3O4

Ag-SiO2@Fe3O4

Fig. 5 – Hysteresis loops of Fe3O4, SiO2@Fe3O4 and Ag-SiO2@Fe3O4 nanoparticles.

115J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

116 J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

to the blocking effect by amorphous silica shell. As it wasexpected, the adsorbents possessed the superparamagneticcharacterization due to the minimized coercive force (tend tozero) and a negligible magnetization hysteresis. Therefore,the superparamagnetic adsorbents could be separated fromfly ashes when used in real flue gas.

298 296 294 292 290 288 286 284 282 280

Cou

nts

(s)

Binding Energy (eV)

C1s284.8eVa

740 735 730 725 720 715 710 705

Cou

nts

(s)

Binding Energy (eV)

Fe2pFe2p3/2Fe2pc

380 375 370 365

Cou

nts

(s)

Ag3d

Binding Energy (eV)

Ag3d5/2

Ag3d3/2

e

Fig. 6 – Survey XPS spectra (f) of the prepared Ag-SiO2@Fe3O4 andSi 2p (d) and Ag 3d (e) peaks of hierarchical Ag-SiO2@Fe3O4 magn

2.1.5. XPSThe XPS spectra (Fig. 6) were detected to further explore thesurface composition of themagnetic particles. As illustrated inFig. 6 (f), peaks with binding energy of 284.8, 532.7, 712.4, 103.4and368.2 eVwereattributed toC1s,O1s, Fe2p, Si 2p, andAg3d,respectively (Yang et al., 2011c). It indicated that the elements

544 542 540 538 536 534 532 530 528 526

Cou

nts

(s)

Binding Energy (eV)

O1s532.7eVb

110 108 106 104 102 100 98 96

Cou

nts

(s)

Binding Energy (eV)

Si 2p103.4eVd

1200 1000 800 600 400 200 0

Si2p

C1s

Binding Energy (eV)

Fe2p

O1s

Ag3d

Cou

nts

(s)

f

enlarged areas corresponding to the C1s (a), O1s (b), Fe 2p (c),etic particles.

117J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

C, O, Si, Fe and Ag consisted in the prepared materials. Also,high-resolution spectra were tested to learn the electronicstates of the elements. As shown in Fig. 6 (a), only one obviouspeak with the binding energy of 284.8 eV was observed as C–Cbond, and Fig. 6 (b) showed the O 1s spectrumwith the bindingenergy of 532.7 eV. It could be seen from Fig. 6 (c) that thebindingenergyof Fe2p1/2appearedat 725 eVwhile Fe2p3/2at712 eV,whichwereconsistentwiththereportedvaluesofFe3O4

in the literature (Choi et al., 2017). Additionally, the bindingenergy of 103.4 eV was attributed to the Si–O bond, and thespectrum was displayed in Fig. 6 (d). In the Fig. 6 (e), the twopeakswere correspondinglywith the bindingenergyofAg3d 3/2 (374.6 eV) and Ag 3d 5/2 (368.6 eV). The values were inaccordance with the results of XRD and FT-IR.

2.1.6. BETThe nitrogen adsorption–desorption isotherms of the synthe-sized materials are shown in Appendix A Fig. S1, and thesurface areas and physical properties calculated from theseisotherms are summarized in Appendix A Table S1. Theisothermexhibited a typeH1 hysteresis at high relative pressure(Cao et al., 2017). The surface area and pore volume of thesynthesized SiO2@Fe3O4 were 15.821 m2/g‐ and 3.6350 cm3/g‐,respectively. Moreover, the average pore size of SiO2@Fe3O4 was14.753 nm. After the loading of Ag nanoparticles, there was asmall decrease in surface area and pore volume but a slightincrease in average pore size, indicating that the small porechannels were blocked (Cao et al., 2017; Han et al., 2012).

2.2. Hg0 removal performances of As-prepared materials

Hg0 removal performances were detected in the bench-scaleexperimental facility under the gas condition of N2 + 10% O2.The Hg0 breakthrough curves of As-prepared materials areshown in Fig. 7. Fe3O4 microspheres had the poorest activityfor Hg0 capture, and the Hg0 removal efficiency was only 20%.SiO2@Fe3O4 possessed a poor activity for Hg0 removal with anefficiency of 45% in the initial few minutes and finallydropped to about 30%. However, the Hg0 removal efficiencycould approach as high as 92% over Ag-SiO2@Fe3O4

0 20 40 60 80 100 1200

20

40

60

80

100

SiO2@Fe3O4

Ag-SiO2@Fe3O4

Fe3O4C/C

0 (%

)

Reaction Time (min)

Fig. 7 – Hg0 breakthrough curves over Fe3O4, SiO2@Fe3O4 andAg-SiO2@Fe3O4.

nanoparticles. The initial inlet Hg0 concentration was 100 μg/m3 in flue gas, and the adsorption capacity was calculated in120 min reaction. The Hg0 adsorption capacities of Fe3O4,SiO2@Fe3O4, Ag-SiO2@Fe3O4 were 30.97, 40.53, 116 μg/g in120 min at 150°C, respectively. It indicated that adsorptioncapacity of hierarchical magnetic composites was muchhigher than that of pure Fe3O4 and SiO2@Fe3O4.

The Hg0 removal efficiencies of the prepared hierarchicalAg-SiO2@Fe3O4magnetic nanoparticleswere testedat differenttemperatures under the gas condition of N2 + 10% O2, and theresults are displayed in Fig. 8. As expected, the preparedadsorbents have an excellent removal performance at lowtemperature (lower than that at 150°C), while the Hg0 removalefficiency at 200°C was only 58%. At 50°C, the adsorbent had aremarkable performance, with the removal efficiency as about90% in the initial few minutes and then got to the maximumefficiency of 92% after 30 min reaction. However, after reactionfor 80 min, the Hg0 removal efficiency dropped to lower than80%. At 150°C, theHg0 removal efficiencywas approach to 92%,andkepthigher than85%even100 minreaction.Obviously,Ag-SiO2@Fe3O4 adsorbent has the best Hg0 removal efficiencyamong all the As-preparedmaterials at 150°C.

To investigate the effects of different gas components onHg0 removal over Ag-SiO2@Fe3O4, O2 and SO2 were selected asprimary gas components, and the results are shown in Fig. 9.The reaction temperature was set at 150°C, and the reactiontimewas120 min.As illustrated inFig. 8, under theconditionofN2 and 10% O2, the Hg0 removal efficiency was approximately84% during the 120 min. When the adsorbents reacted underpure N2, the Hg0 removal efficiency sharply decreased to 32%.Combining with Hg0, the atomic oxygen generated by catalysisof Fe3O4 formed the HgO on the surface of adsorbent. Thisindicated that oxygen was beneficial for the Hg0 removalreaction, which may be attributed to the atomic oxygengenerated by catalysis of Fe3O4 and Ag, and then active Ocombined with Hg0 were transformed to HgO on the surface ofAg-SiO2@Fe3O4 particles. The literature reported the Reaction(2) which could illustrate the catalysis of Fe3O4 for O2 to formO(Yang et al., 2011a). According to the literatures, surfacechemisorbed atomic oxygen was easily formed on the Ag

0 20 40 60 80 100 1200

20

40

60

80

100

Rem

oval

Effi

cien

cy (%

)

Reaction Time (min)

50 oC

100 oC

150 oC

200 oC

Fig. 8 – Hg0 removal performance of Ag-SiO2@Fe3O4 undervarious temperatures.

0

10

20

30

40

50

60

70

80

90

100pure N2

N2 +10% O2

N2 +10% O2 +100ppmSO2

N2 + 5% O2 + 100ppmSO2

N2 + 5% O2 + 200ppmSO2

N2 + 5% O2 + 500ppmSO2

Hg 0

Rem

oval

Eff

icie

ncy

(%)

Fig. 9 – Effects of different gas components on Hg0 removalover Ag-SiO2@Fe3O4.

0 20 40 60 80 100 120

Mer

cury

Sig

nal

Time (min)

500

200

300

400

Tem

pera

ture

(°C

)

600

Fig. 10 –MercurydesorptionperformancesoverAg-SiO2@Fe3O4.

118 J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

nanomaterials (Yang et al., 2011b;Ma et al., 2014; Kaichev et al.,2003). The synergetic interaction might exist between Ag andFe3O4 for generating atomic oxygen, which then contributed tothe Hg0 removal.

Fe2þ þ 1=4 O2→Fe3þ þ 1=2 �O ð2Þ

Although amalgam was the primary adsorption mecha-nism for Hg0 adsorption, such process was also beneficial forenlarging the mercury adsorption capacities.

When 100 ppm SO2 was added into the simulated flue gas,the Hg0 removal efficiency declined to 35%. It indicated thatSO2 has a poison effect on mercury surface uptake. The SO2

reacted with silver atoms to form Ag2S on the surface of Ag-SiO2@Fe3O4 magnetic nanoparticles, whichmay interfere withthe chemical adsorption of Hg0 from flue gas. However, theHg0 removal efficiency was improved when the SO2 concen-tration reached 500 ppm in flue gas, the Hg0 removalefficiency increased to 60%. Previous studies postulated amechanism for the heterogeneous uptake and oxidization ofSO2 on iron oxides (Yang et al., 2011b; Fu et al., 2007), and thereactions can be described as Eqs. (3) and (4):

≡ FeIII−OHþ SO2 gð Þ→ ≡ FeIIIOSO2− þHþ ð3Þ

≡ FeIIIOSO2−→ ≡ FeII þ SO3•− ð4Þ

HgOþ SO3•−→HgSO4 ð5Þ

HgSO4 þHg0→Hg2SO4 ð6Þ

As shown in Eqs. (3) and (4), the uptake of SO2 on ironoxides may involve hydroxyl groups on the surface. Inabsence of hydroxyl groups, the uptake of SO2 on ironoxides was negligible (Yang et al., 2011b; Fu et al., 2007).Then HgO on the surface of Ag-SiO2@Fe3O4 may transformto Hg2SO4 and HgSO4 in the presence of SO3•− formed inReactions (5) and (6). Furthermore, mercurous sulfate waspreviously observed as a mercury product in a photochem-ical oxidation process for elemental mercury oxidation andcapture (McLarnon et al., 2005; Granite and Pennline, 2002).Therefore, elemental mercury capture by Ag-SiO2@Fe3O4

was obviously interfered by the high concentration of SO2.

As a result, the presence of the high concentration of SO2

performed positive effect on elemental mercury capture byAg-SiO2@Fe3O4 at 150°C.

Hg-temperature programmed desorption (Hg-TPD) wascarried out at the lab-scale fixed-bed adsorption systemunder the gas condition of N2, and the results are shown inFig. 10. The initial temperature of the TPD system was 150°C,and then the temperature increased with a heating rate of5°C/min. When the temperature reached 500°C, it maintainedfor a period of time until the adsorbed Hg0 desorbedcompletely. As shown in Fig. 10, the peak at the time of75 min was the desorption peak of Ag–Hg amalgam, whileanother one at the time of 85 min was the desorption peak ofHg–O on the Fe3O4 particles. In the process of Hg0 adsorptionunder O2 condition, atomic oxygen was generated by catalysisof Fe3O4, and the atomic oxygen combing with Hg0 formed theHgO on the surface of Fe3O4 particles. After desorption forabout 2 hr, the desorption profile became flat, indicating thatthe adsorbed mercury was mostly released from the adsorp-tion materials.

Furthermore, the reusability tests of Ag-SiO2@Fe3O4 wereinvestigated, and the results were shown in Fig. S2. Theadsorbed Hg0 could be released from the surface of Ag-SiO2@Fe3O4 composite through TPD method. As shown in AppendixA Fig. S2, six cycles of Hg0 adsorption and desorption testswere carried out. The Hg0 removal efficiency was almost thesame as 90% in the first three cycles, and had a slight decreaseafter the fourth cycle. The results indicated that the synthe-sized Ag-SiO2@Fe3O4 adsorbents were promising adsorbentsfor Hg0 removal due to its excellent reusability.

2.3. Hg0 removal mechanism over Ag-SiO2@Fe3O4

On the basis of Hg0 removal performances and characteriza-tion results of Ag-SiO2@Fe3O4, it could be concluded that themagnetic particles Ag-SiO2@Fe3O4 performed the best adsorp-tion performance at the temperature of 150°C. In thesynthesis process of Ag-SiO2@Fe3O4, atomic silver was suc-cessfully loaded on the material surface. Ag nanoparticleswhich supplied active sites for generating Ag–Hg amalgamplayed an important role in Hg0 adsorption. Moreover,

119J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

dispersive Fe3O4 micro-particles avoided the agglomeration ofAg nanoparticles. In addition, mesoporous silica providedsufficient active room for reactants due to its large specificsurface area. The diffusion ability of silica core-shell made theAg nanoparticles diffusion rather than agglomeration, whichensured sufficient Ag active sites for Hg0 removal.

It was observed that Hg0 was adsorbed from flue gaswithout SO2 by Ag to form the silver amalgam at low temperaterange (about 50–150°C). Moreover, atomic oxygen was gener-ated by catalysis of Fe3O4, and then formed the HgO on thesurface of Fe3O4 particles combining with Hg0. When SO2 wasin flue gas, low concentration of SO2 competed the silver activesites with Hg0 and formed Ag2S on the surface of Ag-SiO2@Fe3O4 magnetic nanoparticles, leading to the reduction of Hg0

removal efficiency. However, the high concentration of SO2 influe gas could be oxidized by iron oxides to SO3•−, and then theHgO on the surface of Ag-SiO2@Fe3O4 may transform to Hg2SO4

and HgSO4 in the presence of SO3•−, which was beneficial forthe conversion and adsorption of Hg0.

3. Conclusion

In this study, hierarchical Ag-SiO2@Fe3O4 magnetic particleswere synthesized for Hg0 removal from simulated non-ferroussmelting flue gas. The characterizations results showed thatSiO2 was coated on the surface of Fe3O4 micro-particles andformed SiO2@Fe3O4 core-shell structure. Ag-SiO2@Fe3O4 washomogeneous spherical morphology with the diameter of260 ± 10 nm. The hysteresis loops of the adsorbents revealedhigh magnetization, indicating the separation performancefrom fly ash.

The hierarchical Ag-SiO2@Fe3O4 magnetic particles exhib-ited prominent Hg0 removal efficiency as approximately 92%at 150°C. For Hg0 removal process, it was ascribed to achemical-adsorption process. Hg0 was adsorbed by atomicAg to form the silver amalgam, and atomic oxygen generatedby catalysis of Fe3O4 was transformed to HgO on the surface ofFe3O4 particles combining with Hg0. With SO2 in flue gas, SO2

could occupy the silver active sites and form Ag2S on thesurface of Ag-SiO2@Fe3O4 magnetic nanoparticles, leading tothe reduction of Hg0 removal efficiency. However, the highconcentration of SO2 in flue gas could be oxidized by ironoxides to SO3•−, and then the SO3•− combined with HgO on thesurface of Ag-SiO2@Fe3O4 were transformed to Hg2SO4 andHgSO4. This was beneficial to the conversion and adsorptionof Hg0. Moreover, the magnetic particles could be easilydesorbed via Hg-TPD system at 500°C.

In conclusion, Ag-SiO2@Fe3O4 magnetic particles have anexcellent adsorption performance for Hg0 at low temperature.It could realize cyclic utilization via magnetically separationand desorption regeneration, which enabled the reduction ofapplication cost and secondary pollution.

Acknowledgments

This study was supported by the National Key R&D Program ofChina (No. 2017YFC0210500), the National Natural Science

Foundation of China (No. 51508525), the Key Research andDevelopment Program of Ningxia Hui Autonomous Region(No. 2016KJHM31).

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.jes.2018.11.014.

R E F E R E N C E S

Cao, T.T., Li, Z., Xiong, Y., Yang, Y., Xu, S.M., Bisson, T., et al., 2017.Silica-silver nanocomposites as regenerable sorbents for Hg0

removal from flue gases. Environ. Sci. Technol. 51,11909–11917.

Choi, K.H., Park, S.Y., Park, B.J., Jung, J.S., 2017. Recyclable Ag-coated Fe3O4@TiO2 for efficient photocatalytic oxidation ofchlorophenol. Surf. Coat. Technol. 320, 240–245.

Dong, L., Xie, J.K., Fan, G., Huang, Y., Zhou, J., Sun, Q.K., et al., 2017.Experimental and theoretical analysis of element mercuryadsorption on Fe3O4/Ag composites. Korean J. Chem. Eng. 34,2861–2869.

Du, G.H., Liu, Z.L., Xia, X., Chu, Q., Zhang, S.M., 2006. Character-ization and application of Fe3O4/SiO2 nanocomposites. J. Sol-Gel Sci. Techn. 39, 285–291.

Du, J.Y., Liu, J.F., Gong, P.W., Tian, M., Sun, L., Ji, S.J., et al., 2017.Construction of a novel fluorinated graphene-based magneticnanocomposite and its application in cancer photo-chemotherapy. Mater. Lett. 196, 165–167.

Fu, H.B., Wang, X., Wu, H.B., Yin, Y., Chen, J.M., 2007. Heteroge-neous uptake and oxidation of SO2 on iron oxides. J. Phys.Chem. C 111, 6077–6085.

Granite, E.J., Pennline, H.W., 2002. Photochemical removal ofmercury from flue gas. Ind. Eng. Chem. Res. 41, 5470–5476.

Han, J., Fang, P., Jiang, W., Li, L., Guo, R., 2012. Ag-nanoparticle-loaded mesoporous silica: spontaneous formation of Agnanoparticles and mesoporous silica SBA-15 by a one-potstrategy and their catalytic applications. Langmuir : the ACSjournal of surfaces and colloids 28, 4768–4775.

Huang, C.K., Hou, C.H., Chen, C.C., Tsai, Y.L., Chang, L.M., Wei, H.S., et al., 2008. Magnetic SiO2/Fe3O4 colloidal crystals. Nano-technology 19, 055701.

Kaichev, V.V., Bukhtiyarov, V.I., Hävecker, M., Knop-Gercke, A.,Mayer, R.W., Schlögl, R., 2003. The nature of electrophilic andnucleophilic oxygen adsorbed on silver. Kinet. Catal. 44,432–440.

Liang, K., Qiao, X.J., Sun, Z.G., Guo, X.D., Wei, L., Qu, Y., 2017.Preparation and microwave absorbing properties of grapheneoxides/ferrite composites. Appl. Phy. A 123, 445.

Liu, W., Vidić, R.D., Brown, T.D., 1998. Optimization of sulfurimpregnation protocol for fixed-bed application of activatedcarbon-based sorbents for gas-phase mercury removal. Envi-ron. Sci. Technol. 32, 531–538.

Liu, Z.L., Peng, B., Chai, L.Y., Liu, H., Yang, S., Yang, B.T., et al.,2017a. Selective removal of elemental mercury from high-concentration SO2 flue gas by thiourea solution and investi-gation of mechanism. Ind. Eng. Chem. Res. 56, 4281–4287.

Liu, Z.L., Wang, D.L., Peng, B., Chai, L.Y., Liu, H., Yang, S., et al.,2017b. Transport and transformation of mercury during wetflue gas cleaning process of nonferrous metal smelting.Environ. Sci. Poll. Res. 24, 22494–22502.

Liu, Z.L., Wang, D.L., Peng, B., Chai, L.Y., Yang, S., Liu, C., et al.,2017c. Mercury re-emission in the smelting flue gas cleaningprocess: the influence of arsenite. Energ. Fuel. 31, 11053–11059.

120 J O U R N A L O F E N V I R O N M E N T A L S C I E N C E S 7 9 ( 2 0 1 9 ) 1 1 1 – 1 2 0

Louie, D., 2005. Handbook of Sulphuric acid manufacturing. DKLEngineering. Inc, Thornhill, Ontario, Canada.

Ma, L., Wang, D.S., Li, J.H., Bai, B.Y., Fu, L.X., Li, Y.D., 2014. Ag/CeO2

nanospheres: efficient catalysts for formaldehyde oxidation.Appl. Catal. B 148, 36–43.

Ma, Y.P., Xu, H.M., Qu, Z., Yan, N.Q., Wang, W.H., 2015. Removal ofmercury from flue gas from nonferrous metal smelting, by useof mercury chloride solution, and mechanisms of inhibition bysulfur dioxide. Res. Chem. Intermediat. 41, 5889–5905.

Ma, Y.P., Mu, B.L., Yuan, D.L., Zhang, H.Z., Xu, H.M., 2017. Design ofMnO2/CeO2-MnO2 hierarchical binary oxides for elementalmercury removal from coal-fired flue gas. J. Hazard. Mater. 333,186–193.

McLarnon, C.R., Granite, E.J., Pennline, H.W., 2005. The PCOprocess for photochemical removal of mercury from flue gas.Fuel Process. Technol. 87, 85–89.

Pacyna, E.G., Pacyna, J., Sundseth, K., Munthe, J., Kindbom, K.,Wilson, S., et al., 2010. Global emission of mercury to theatmosphere from anthropogenic sources in 2005 and projec-tions to 2020. Atmos. Environ. 44, 2487–2499.

Pirrone, N., Cinnirella, S., Feng, X.B., Finkelman, R., Friedli, H.,Leaner, J., Stracher, G., Streets, D., et al., 2010. Global mercuryemissions to the atmosphere from anthropogenic and naturalsources. Atmos. Chem. Phys. 10, 5951–5964.

Selin, H., 2014a. Global environmental law and treaty-making onhazardous substances: the Minamata Convention and mer-cury abatement. Global Environ. Polit. 14, 1–19.

Selin, N.E., 2014b. Global change and mercury cycling: challengesfor implementing a global mercury treaty. Environ. Toxicol.Chem. 33, 1202–1210.

Streets, D.G., Hao, J.M., Wu, Y., Jiang, J.K., Chan, M., Tian, H.Z, et al.,2005. Anthropogenic mercury emissions in China. Atmos.Environ. 39, 7789–7806.

Tomer, V.K., Adhyapak, P.V., Duhan, S., Mulla, I.S., 2014. Humiditysensing properties of Ag-loaded mesoporous silica SBA-15nanocomposites prepared via hydrothermal process.Micropor. Mesopor. Mat. 197, 140–147.

Wang, J.W., Yang, J.L., Liu, Z.Y., 2010a. Gas-phase elementalmercury capture by a V2O5/AC catalyst. Fuel Process. Technol.91, 676–680.

Wang, S.X., Song, J.X., Li,G.H.,Wu,Y., Zhang, L.,Wan,Q., et al., 2010b.Estimatingmercury emissions from a zinc smelter in relation toChina's mercury control policies. Environ. Poll. 158, 3347–3353.

Wu, Q.R., Wang, S.X., Zhang, L., Song, J.X., Yang, H., Meng, Y., 2012.Update of mercury emissions from China's primary zinc, leadand copper smelters, 2000-2010. Atmos. Chem. Phys. 12,11153–11163.

Xiao, Y., Liu, Q., Liu, Z.Y., Huang, Z.G., Guo, Y.X., Yang, J.L., 2008.Roles of lattice oxygen in V2O5 and activated coke in SO2

removal over coke-supported V2O5 catalysts. Appl. Catal. B-Environ. 82, 114–119.

Xu, H.M., Qu, Z., Huang, W.J., Mei, J., Chen, W.M., Zhao, S.J, et al.,2015a. Regenerable Ag/graphene sorbent for elemental mer-cury capture at ambient temperature. Colloid. Surface A 476,83–89.

Xu, H.M., Qu, Z., Zong, C.X., Huang, W.J., Quan, F.Q., Yan, N.Q.,2015b. MnOx/graphene for the catalytic oxidation and adsorp-tion of elemental mercury. Environ. Sci. Technol. 49,6823–6830.

Yang, S.J., Guo, Y.F., Yan, N.Q., Qu, Z., Xie, J.K., Yang, C., Jia, J.P.,2011a. Capture of gaseous elemental mercury from flue gasusing a magnetic and sulfur poisoning resistant sorbentMn/γ-Fe2O3 at lower temperatures. J. Hazard. Mater. 186,508–515.

Yang, S.J., Guo, Y.F., Yan, N.Q., Wu, D.Q., He, H.P., Xie, J.K., et al.,2011b. Remarkable effect of the incorporation of titanium onthe catalytic activity and SO2 poisoning resistance of magneticMn–Fe spinel for elemental mercury capture. Appl. Catal. B-Environ. 101, 698–708.

Yang, S.J., Yan, N.Q., Guo, Y.F., Wu, D.Q., He, H.P., Qu, Z., et al.,2011c. Gaseous elemental mercury capture from flue gas usingmagnetic nanosized (Fe3-xMnx)1-δO4. Environ. Sci. Technol.45, 1540–1546.

Zhang, Y.X., Yu, X.Y., Jin, Z., Jia, Y., Xu, W.H., Luo, T., et al., 2011.Ultra high adsorption capacity of fried egg jellyfish-like γ-AlOOH(Boehmite)@SiO2/Fe3O4 porous magnetic microspheresfor aqueous Pb (II) removal. J. Mater. Chem. 21, 16550–16557.

Zhang, L., Wang, S., Wu, Q.R., Meng, Y., Yang, H., Wang, F.Y, etal., 2012. Were mercury emission factors for Chinese non-ferrous metal smelters overestimated? Evidence fromonsite measurements in six smelters. Environ. Poll. 171,109–117.

Zhang, L., Wang, S.X., Wang, L., Wu, Y., Duan, L., Wu, Q.R., et al.,2015. Updated emission inventories for speciated atmosphericmercury from anthropogenic sources in China. Environ. Sci.Technol. 49, 3185–3194.