Combustion and Flame 191 (2018) 9–18

Contents lists available at ScienceDirect

Combustion and Flame

journal homepage: www.elsevier.com/locate/combustflame

Chemical-looping combustion of plastic wastes for in situ inhibition of

dioxins

Haibo Zhao

∗, Jinxing Wang

State Key Laboratory of Coal Combustion, School of Energy and Power Engineering, Huazhong University of Science and Technology, Wuhan 430074, PR

China

a r t i c l e i n f o

Article history:

Received 24 March 2017

Revised 5 October 2017

Accepted 22 December 2017

Keywords:

Chemical looping combustion

Waste management

PCDD/Fs

Dechlorination

Chlorine substitution

a b s t r a c t

Polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs) as toxic by-products are inevitably

emitted during conventional air incineration processes of chlorine containing wastes (typically, plastic

wastes). The presence of O 2 in the conventional waste incineration not only participates in carbon gasifi-

cation and rearrangement as an oxygen source during the de novo synthesis process of PCDD/Fs, but also

promotes chlorination through generating more active Cl 2 via Deacon reaction. Chemical looping com-

bustion (CLC), which creates an O 2 -free atmosphere in the fuel reactor, was proposed to dispose plastic

wastes. Comparative experiments [conventional PW incineration vs. in situ gasification - chemical looping

combustion ( i G-CLC) using CaO-decorated Fe 2 O 3 /Al 2 O 3 as oxygen carrier] were conducted to measure the

distribution properties of 17 toxic PCDD/Fs congeners. The total amount and toxic equivalency quantity of

PCDD/Fs have been reduced by 94 % and 89 %, respectively, in i G-CLC. The absence of O 2 in fuel reactor

and lower Cl 2 yield (due to the restriction of Deacon reaction and effective dechlorination by CaO) lead

to the significant inhibition of the PCDD/Fs formation. Chlorine substitution modelling also demonstrated

that the chlorine substitution probabilities for the formation of 7 toxic congeners of PCDDs and 10 toxic

congeners of PCDFs are significantly reduced in i G-CLC.

© 2017 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1

m

e

h

d

i

a

c

o

c

p

p

c

i

a

p

t

d

t

a

m

n

z

p

r

t

t

a

e

d

a

s

t

o

s

w

h

0

. Introduction

A fast industrial development has led to increased amounts of

unicipal solid waste (MSW) [1] . These solid wastes, if not prop-

rly disposed, are serious threats to the environment and public

ealth [2–4] . On the other hand, MSW is potentially valuable fuel

ue to its high calorific value and embodied energy. Among the ex-

sting waste treatment technologies, incineration has been broadly

dopted [5] . However, serious attention has been paid to poly-

hlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) as sec-

ndary contaminations, especially for the incineration of chlorine

ontaining polymers such as non-renewable and non-recyclable

lastic wastes (PWs) [6] . Consequently, the problem of MSW dis-

osal urgently needs to be addressed and solved.

To date, it has been verified that many factors (such as O 2 con-

entration, chlorine source, carbon residue and catalytic metal ox-

de) could affect the formation of PCDD/Fs in the waste inciner-

tion processes [7–10] . In particular, chlorine and oxygen sources

lay an important role in the formation of PCDD/Fs [7–9] . Al-

hough the PCDD/Fs formation pathways have not been fully un-

∗ Corresponding author.

E-mail addresses: klinsmannzhb@163.com , hzhao@mail.hust.edu.cn (H. Zhao).

r

e

h

ttps://doi.org/10.1016/j.combustflame.2017.12.026

010-2180/© 2017 The Combustion Institute. Published by Elsevier Inc. All rights reserved

erstood, it is generally believed that precursor conversion (high-

emperature homogeneous reactions or low-temperature catalytic-

ssisted heterogeneous reactions) and de novo synthesis are two

ajor mechanisms [7–9] . In the precursor conversion mecha-

ism, organic molecules combine to form PCDD/Fs. Chloroben-

enes (CBzs), chlorophenols (CPs), polychlorobenzenes (PCBzs),

olychlorophenols (PCPs), etc. have been considered as main chlo-

inated precursors [7–10] . The de novo synthesis is involved with

he breakdown of a carbonaceous matrix with simultaneous oxida-

ion and chlorination under the influence of metal catalysts (such

s copper) [7–9] .

In recent decades, effective means for controlling PCDD/Fs

mission and meeting stringent emission regulations have been

eveloped, including activated carbon injection (ACI) [11] , and the

ddition of suppressants such as calcium-based additives [12] and

ulphur-containing compounds [13] . However, ACI technology only

ransfers gaseous PCDD/Fs to the solid phase (in fly ash), with-

ut reducing the total emission of PCDD/Fs [14] . The addition of

uppressants will increase operation cost. The most cost-effective

ay of reducing PCDD/Fs emission is to minimize their generation

ather than to destroy or collect them after they have been gen-

rated. Therefore, it is essential to explore novel approaches to in-

ibit their formation in-situ , based on their formation mechanisms.

.

10 H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18

Fig. 1. Schematic description of chemical looping combustion technology.

e

g

a

o

s

t

t

p

i

f

s

u

i

c

[

w

[

a

p

e

d

o

c

p

t

P

d

f

o

v

s

s

t

c

a

t

e

s

h

t

a

p

t

m

F

c

c

a

2

2

m

c

p

s

d

%

p

c

a

a

It is widely believed that O 2 participates in carbon gasifica-

tion and rearrangement as an oxygen source during the de novo

synthesis process of PCDD/Fs [7–9] . Therefore, O 2 -deficient con-

dition is very beneficial to inhibit the formation of PCDD/Fs via

the de novo synthesis route. On the other side, the absence of O 2

will effectively suppress the metal-catalyzed Deacon reaction (R1)

4HCl + O 2 ↔ 2Cl 2 + 2H 2 O [15] . As known, Cl 2 has a higher activity

for chlorination than HCl, and the chlorination of the benzene ring

by Cl 2 is an essential step in both precursor conversion and de novo

synthesis [7–9] . Therefore, a low Cl 2 amount (which is due to the

O 2 -free condition) is expected to help the inhibition of PCDD/Fs. To

sum up, creating an O 2 -free atmosphere during the combustion of

chlorine containing solid wastes may be very advantageous from

the perspective of PCDD/Fs inhibition.

Chemical-looping combustion (CLC) [16] , which has been

emerging as a promising next-generation CO 2 capture technol-

ogy with the advantages of inherent CO 2 separation, cascaded

utilization of energy and low NOx [17] , utilizes an oxygen

carrier (OC, usually a metal oxide) to transport oxygen from

the air to the fuel, avoiding direct reaction between fuel (PW

here) and air and, indeed, creating an O 2 -free atmosphere for

fuel combustion. We [18] therefore proposed CLC as a promis-

ing waste management technology for in situ inhibition of

PCDD/Fs. The designed CLC prototype usually consists of two

separate reactors: an air reactor (AR) and a fuel reactor (FR)

[19] , as shown in Fig. 1 . Here Me x O y denotes the OC in the

oxidized form and Me x O y-1 in the reduced form. In the FR, fuel

will react with Me x O y to generate CO 2 and H 2 O via reaction

(R2) C n H m

+ (2 n + m )Me x O y → n CO 2 + m H 2 O + (2 n + m )Me x O y -1 .

In the AR, the reduced OC, Me x O y-1 , is re-oxidized to its

originally oxidized form by air according to reaction (R3)

O 2 + 2Me x O y -1 → 2Me x O y .

Bi et al. [20] measured the formation characteristics of PCDD/Fs

in the CLC of polyvinyl chloride (PVC) pyrolysis gas using CaSO 4 as

OC in a two-stage reactor, where PVC was pyrolyzed in N 2 atmo-

sphere in the 1st reactor and then the pyrolysis gas was carried

by N 2 to the 2nd reactor to react with CaSO 4 . Although PCDD/Fs

or their precursors were generated in the 1st pyrolysis reactor,

the yield and International Toxicity Equivalent Quantity (I-TEQ) of

PCDD/Fs after the CLC reactor were significantly lower than those

in air incineration, which is attributed to the absence of O . How-

2

ver, the syngas/pyrogas-CLC technique requires another separate

asification/pyrolysis reactor, which will increase the complexity

nd operation cost of CLC system. Furthermore, using CaSO 4 as

xygen carrier may suffer from SO x emission and particle abra-

ion/fragmentation [21] . Considering plastic wastes [usually con-

aining a very high volatile content (e.g., 93%)] are quite easier

o be pyrolyzed or gasified than other solid fuels (e.g . , coal char,

etroleum coke, or even biomass char), we first proposed to utilize

n situ gasification-CLC ( i G-CLC, where the pyrolysis/gasification of

uel and the reduction of combustible gases by OC occur in the

ame reactor) to dispose PW [22,23] . The performance of i G-CLC

sing PW as fuel and iron oxide as oxygen carrier was evaluated

n a batch-operated fluidized bed reactor, and very high carbon

onversion (up to 98%) and CO 2 yield (up to 97%) can be attained

22,23] . Further dechlorination through adsorbent-decorated OC

as examined, and the dechlorination efficiency can reach ca. 80%

18] , which will further constrain the presence of Cl 2 and other

ctive chlorine sources. We [23] also measured the organic com-

ounds in the exhaust gases during the i G-CLC of PWs, and discov-

red that utilizing CaO-decorated iron ore as OC can effectively re-

uce the emission of chlorobenzene (as an indicative intermediate

f PCDD/Fs), which is attributed to the effective dechlorination via

hlorine adsorbent ( i.e., CaO). Although these encouraging results

rovided indirect evidence to in-situ inhibition of PCDD/Fs by CLC,

o the best of our knowledge, the distribution properties of 17 toxic

CDD/Fs congeners as the direct evidence have not been measured

uring the i G-CLC process of PWs to date. More importantly, the

ormation and inhibition mechanisms of PCDD/Fs in i G-CLC based

n the experimental measurement have not been investigated pre-

iously.

In order to examine the feasibility of the CLC technique for in

itu inhibiting the formation of PCDD/Fs, the following questions

hould be addressed. ( 1 ) How are 17 toxic PCDD/Fs congeners dis-

ributed in the i G-CLC process, especially when compared to the

onventional PW incineration? ( 2 ) How are the PCDD/Fs formed

nd inhibited in i G-CLC, respectively? To address the first ques-

ion, the 17 toxic PCDD/Fs congeners from conventional PW incin-

ration and i G-CLC using CaO-decorated Fe 2 O 3 /Al 2 O 3 as OC were

ampled and quantified by a high-resolution gas chromatography/

igh-resolution mass spectrometer (HRGC/HRMS). Then, in order

o explore the formation and inhibition mechanisms of PCDDs

nd PCDFs in i G-CLC, a chlorine substitution model was pro-

osed to calculate the chlorine substitution probability based on

he PCDD/Fs measurements. Also, the Cl 2 yield measurement via

ethyl orange spectrophotometry, the organic matter analysis via

ourier-transform mass spectrometry, as well as particle (oxygen

arrier and fly ash) characterization via ESEM-EDX and XPS were

onducted to help understand the formation and inhibition mech-

nisms of PCDDs and PCDFs in i G-CLC.

. Experimental

.1. Materials

A medical perfusion tube (which is made of polyolefin ther-

oplastic elastomer. Polyethylene and polybutylene are the main

omponents, and a small amount of chlorine is also added to im-

rove its performance) was selected for this study as a typical PW

ample. The ultimate analysis of PW shows that the contents (wt%,

.a.f.) of C, Cl, O, S, N and H are 73.89 %, 4.92 %, 3.39 %, 0.16 %, 0.72

and 10.82 %, respectively. Besides, the proximate analysis com-

onents of PW includes volatile (93.79 wt%), ash (6.1 wt%), fixed

arbon (0.08 wt%) and moisture (0.03 wt%). In addition, the ash

nalysis of PW shows that the contents (wt%) of SiO 2 , Fe 2 O 3 , CaO

nd Al O are 34.92 %, 19.23 %, 19.54 % and 15.63 %, respectively.

2 3

H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18 11

Fig. 2. Overview of the fluidized-bed reactor.

T

t

[

o

6

p

d

c

t

s

t

t

[

F

o

a

O

t

2

a

b

t

g

t

(

u

c

c

i

v

d

a

t

t

a

g

i

o

C

r

t

(

t

t

p

g

m

C

y

d

c

H

g

u

w

i

t

t

i

t

i

O

0

d

w

t

p

3

w

o

F

s

d

c

r

f

r

s

m

v

[

i

H

C

m

I

T

φ

w

f

r

n

n

h

O

t

a

a

a

b

2

w

he detailed procedures for the ultimate and proximate analyses of

he PW can been found in our previous publications [18,22,23] .

The OC is one of important components in the CLC operation

24,25] . Iron oxide demonstrates a competitive superiority because

f its low price [26,27] . Following our previous research [18,22] ,

0 wt% Fe 2 O 3 /40 wt% Al 2 O 3 , which was synthesized using the co-

recipitation method, was selected as a typical OC. Since in situ

echlorination may suppress the generation of PCDD/Fs [23] , we

onsidered to decorate the chlorine adsorbent into the OC par-

icles [18,22,23] . In our previous publications, the effects of ad-

orbent (K 2 O, Na 2 O, CaO), the decoration method (wet impregna-

ion, physical mixture and coprecipitation), and the loading con-

ent on dechlorination efficiency in CLC have been investigated

18] . This study confirmed that 5 wt% CaO decoration on synthetic

e 2 O 3 /Al 2 O 3 OC via wet impregnation can reach an optimum level

f dechlorination. Therefore, a 5 wt% CaO-decorated Fe 2 O 3 /Al 2 O 3

s OC was used in this study. More details for synthesizing these

C particles can be found in Section S1 of the supplementary ma-

erial (SM).

.2. Apparatus

Comparative combustion experiments (conventional air inciner-

tion and i G-CLC) were conducted in a laboratory-scale fluidized

ed reactor that simulates a real i G-CLC process. An overview of

he fluidized-bed system is shown in Fig. 2 , which includes a

as feeding unit, a fluidized-bed reaction unit, a PCDD/Fs absorp-

ion apparatus (XAD-2 resin of 15 g) or a Cl 2 absorption apparatus

methyl orange absorption liquid of 100 mL) and a gas detection

nit (an on-line gas analyzer, Gasboard-3151). The gas feeding unit

an control the atmosphere in reactor through switching the in-

oming gases (air, nitrogen and steam), which simulates the PW

ncineration reactor and the AR or FR in i G-CLC. The reactor is a

ertical and cylindrical stainless steel tube and its height and inner

iameter are 890 mm and 26 mm, respectively. It was operated at

mbient pressure. An electrical heating system was used for con-

rolling the temperature of the reactor, which was monitored by a

hermocouple at the furnace center. In these experiments, the re-

ction temperature is 1173.15 K and the flow rate of the incoming

ases is 10 mL s −1 . Air is the fluidizing agent for the conventional

ncineration and the oxidation stage of the i G-CLC, while a mixture

f 40 vol% steam and 60 vol% N 2 is the fluidizing agent for the i G-

LC reduction stage. The exhaust gases were led to a filter, XAD-2

esin (or methyl orange absorption liquid), an electric cooler and

he on-line gas analyzer in sequence to extract particulate matter

fly ash and fine OC particles), absorb the PCDD/Fs generated (or

he Cl 2 generated), remove the steam and measure the concentra-

ions of CO, CO 2 , CH 4 , H 2 and O 2 (dynamically recorded by a com-

uter). In order to avoid the condensation of PCDD/Fs, the exhaust

ases before leading into XAD-2 resin were maintained approxi-

ately 473.15 K. For the on-line gas analyzer, the concentrations of

O 2 , CH 4 and CO were determined by nondispersive infrared anal-

sis (NDIR, with an accuracy of 1% FS) and H 2 by thermal gas con-

uctivity (TCD, with an accuracy of 2% FS), while O 2 by electron

apture detector (ECD, with an accuracy of 2% FS). With respect to

Cl, it is easily water-soluble, and is not detectable by the on-line

as analyzer. We introduced the exhaust gas into the NaHCO 3 sat-

rated solution, where HCl was collected. The HCl concentration

as measured by an ion chromatograph (ICS-90) through detect-

ng the chloride ion concentration in solution [18] . In this study,

he concentrations of sulfide and nitride were not measured due

o relatively few contents of S and N from the ultimate analysis.

The experimental procedure in the batch fluidized bed reactor

s described as follows. First, OC particles of 30 g were injected into

he reactor through the hopper, and then, air of 0.6 NL/min was

ntroduced into the reactor from the bottom to adequately oxidize

Cs during the heating process (including holding at 1173.15 K for

.5 h). Next, a mixture of 40 vol% steam and 60 vol% N 2 was intro-

uced into the reactor as the fluidizing agent. Then PWs of 0.084 g

ere fed into the reactor through the hopper in one shot. When

he gas concentrations of CO 2 , CO, CH 4 , H 2 , and O 2 reached ap-

roximately zero, N 2 (10 mL s −1 ) as purge gas was introduced for

00 s to remove residual vapor. Then, these reduced OC particles

ere re-oxidized through switching air as the fluidizing agent. An-

ther CLC redox cycle can be initiated with the same procedure.

or the conventional incineration, OC particles were replaced with

ilica sand and air was invariably served as fluidizing agent. In or-

er to verify the repeatability of experimental data, 10 successive

ycles for each experimental condition were conducted in the same

eactor. In all of experiments, the same mass (0.084 g) of PWs was

ed into the reactor.

Our previous study [22] has demonstrated that 900 °C for

eaction temperature, 2.5 for supply oxygen ratio, and 40 vol.%

team/60 vol.% N 2 for fluidizing agent can reach optimal perfor-

ance of i G-CLC of PWs. It has been shown that adding water

apour will influence the Deacon reaction in the presence of O 2

28] , which is not the case of the i G-CLC experiments. In a real

G-CLC process, the fluidizing agent of the fuel reactor should be

2 O, CO 2 , or their mixture, in order to attain high-concentration

O 2 in flue gas, without the N 2 dilution. Adding water vapour is

ore beneficial for the gasification of carbon residue than CO 2 .

n this work, we still followed the optimizing operation condition.

he supply oxygen ratio, φ, which was defined by

= n O , OC / n O , PW

(1)

here n O,OC is the molar amount of oxygen in the OCs available

or the i G-CLC process, and n O,PW

is the molar amount of oxygen

equired for full combustion of PWs .

O , OC = m OC βF e 2 O 3 / (3 M F e 2 O 3

)(2)

O , PW

= m PW

(0 βN

M N

+

2 βC

M C

+

2 βS

M S

+

βH

2 M H

− βO

M O

− βCl

2 M Cl

)(3)

ere β i is the mass fraction of active component i (Fe 2 O 3 ) in the

Cs or element i (N, C, S, H, O, Cl) in the PW, m OC is the mass of

he oxidized OC, m PW

is the mass of the PW fed into the reactor,

nd M i is the molar mass of specie i . Eq. (3) was based on the

ssumptions that PW is only oxidized to N 2 , CO 2 , SO 2 and H 2 O

nd that HCl is the only form of chlorine-containing reactants (see

elow, the Cl 2 yield in i G-CLC is very low) [22] .

.3. Measurement and characterization

The distribution properties of 17 toxic PCDD/Fs congeners

ere analyzed quantitatively by a HRGC/HRMS (JMS-800D). The

12 H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18

t

m

η

w

b

a

c

l

n

d

fi

f

3

r

l

t

P

o

c

i

p

t

f

o

p

g

t

g

c

d

t

s

g

l

t

m

S

w

P

f

t

t

o

t

o

c

m

4

4

i

t

r

a

clean-up procedures of PCDD/Fs samples followed US EPA method

1613 [29] . First, 1 ng of 13 C 12 -labeled internal standard was added

to the XAD-2 resin, which was then Soxhlet-extracted using

toluene (purity 99.5%; J.T. Baker, USA) of 250 mL for 24 h. Next, the

Soxhlet-extract solution was condensed to 1-2 mL in a rotary evap-

orator, and then, the concentrated solution was gently blown dried

with high purity nitrogen in a centrifugal tube. Subsequently, the

qualities of 17 toxic congeners of PCDD/Fs were detected by the

HRGC/HRMS with a DB-5MS column (60 m × 0.25 mm × 0.25 μm).

The HRGC temperature programmed procedure is described as fol-

lows: the concentrated solution of 1 μl was injected into the DB-

5MS column at 423.15 K with splitless injection mode, and then

the oven temperature was maintained at 423.15 K for 60 s. Next,

the oven temperature was increased to 463.15 K at 0.417 K s −1 and

then increased to 553.15 K at 0.5 K s −1 . Finally, the oven tempera-

ture was held at 553.15 K for 20 min. All of PCDD/Fs analyses were

repeated at least twice.

After the fluidized bed experiments, the Cl 2 yields under two

combustion conditions were gained through detecting the ab-

sorbance of the methyl orange absorption liquid by an UV/VIS

(Lambda 35). The Cl 2 yield was defined as the ratio of Cl 2 quantity detected to the total mass of PW. The methyl orange

absorption liquid was prepared as follows: first, a 1:6 sulfuric

acid solution was prepared from 100 mL concentrated sulfuric

acid ( ρ = 1.84 g/mL) dissolved in 600 mL deionized water; second,

methyl orange of 0.1 g and absolute ethyl alcohol of 20 mL were

mixed with deionized water, resulting in a solution of methyl

orange stock of 10 0 0 mL; third, potassium bromide of 5.0 g, the

methyl orange stock solution of 250 mL and the 1:6 sulfuric acid

solution of 500 mL were mixed with deionized water, resulting in

methyl orange ready liquid of 10 0 0 mL; fourth, the methyl orange

ready liquid of 20 mL was diluted with deionized water, result-

ing in the methyl orange absorption liquid of 100 mL. The stan-

dard line of Cl 2 content was drawn through the linear regression

method as following: first, potassium bromate of 1.9627 g was dis-

solved in 500 mL deionized water and then was diluted 100 times.

Note that the potassium bromate solution of 1 mL was equivalent

to 50.0 g Cl 2 ; second, the methyl orange ready liquid of 20 mL and

the potassium bromate solution of different volumes (0, 0.20, 0.40,

0.80, 1.20, 1.60, 2.00 mL) were diluted with deionized water to gain

a standard solution of 100 mL; third, the absorbance of these stan-

dard liquids at 507–515 nm was detected by the UV/VIS, and deion-

ized water was used as a reference.

2.4. Data evaluation

To date, considerable public attention has been paid to PCDD/Fs

with chlorine substitution in the 2, 3, 7, 8 positions due to their

toxicity [7–9] . According to the International Toxic Equivalency Fac-

tors, ( TEF ) i , of 17 PCDD/Fs congeners (see Table S1 in SM), the Toxic

Equivalency Quantity ( TEQ ) was calculated as:

T EQ =

n =17 ∑

i =1

[ c i × (T EF ) i ] (4)

where c i is the concentration of 2, 3, 7, 8-positions chlorine sub-

stitution for these 17 congeners of PCDD/Fs, which was calculated

as the detection quantity (ng) of HRGC/HRMS divided by the PW

mass (g).

The i G-CLC performance and the OC reactivity in these

laboratory-scale fluidized bed experiments were evaluated in

terms of combustion efficiency ( η). The combustion efficiency η in

the fuel reactor is related to the oxygen demand (which describes

the fraction of oxygen lacking to achieve complete combustion of

these combustible gases produced in the FR) [30] . Here the com-

bustion efficiency was calculated by integrating the molar flow of

he gaseous product species i in the off gas with the oxygen de-

and of PWs:

=

[1 −

(∫ t 1

0

F CO . out d t + 4

∫ t 1

0

F C H 4 . out d t +

∫ t 1

0

F H 2 . out d t

)/ n O , PW

]

×100% (5)

here F i ,out ( i = CO, CH 4 , H 2 ) is the molar flow of the gaseous com-

ustible species i; t 1 is the duration of each test. Here only CO, CH 4

nd H 2 were considered as incomplete combustion products be-

ause the amount of other light hydrocarbons C x H y (x > 2) was be-

ow instrument (on-line gas analyzer) detection limits. It is unde-

iable that the liquid or the solid products may also be generated

uring these combustion processes of PWs; however, this simpli-

cation cannot result in an impermissible error. The relative error

or carbon mass balance was below 1 %.

. Modelling

Predicting PCDD/Fs isomer patterns from these experimental

esults enables an understanding of PCDD/Fs formation that may

ead to preventive measures. In this study, the chlorine substitu-

ion model was proposed to analyze the formation of PCDDs and

CDFs, respectively. The basic idea is: according to the symmetry

f the PCDD or PCDF molecular structure (see Fig. S1 in SM), the

hlorine substitution probability of 1,4,6,9 position for PCDDs ( χ )

s identical, and so is the chlorine substitution probability of 1,9

osition ( α) or 4,6 position ( β) for PCDFs. An important assump-

ion is: higher chlorinated PCDDs or PCDFs isomers are generated

rom lower chlorinated PCDDs or PCDFs isomers step by step. In

ther words, 2378TCDD (no chlorine substitution occurs at 1,4,6,9

osition) can be considered as the parent of other toxic PCDD con-

eners, and 2378TCDF (neither chlorine substitution at 1,9 posi-

ion nor at 4,6 position) as the parent of other toxic PCDF con-

eners. Thus, the proportions of 7 PCDD congeners (or 10 PCDF

ongeners) can be calculated, as listed in Table S2. The modelling

etails can be found in Section S3 of SM. The predicted concen-

rations of PCDD/Fs were compared with the measurements. The

imilarity ( S ) was defined as Eq. (6) to measure the difference de-

ree of predicted values and experimental values [31] . The simi-

arity is the function of chlorine substitution probabilities. The op-

imum chlorine substitution probability was determined once the

aximum similarity was attained.

=

∑

A i B i /

√ ∑

( A i ) 2 ∑

( B i ) 2

(6)

here A i are the predicted concentrations for these 7 congeners of

CDDs or 10 congeners of PCDFs; B i are the experimental values

or these 7 congeners of PCDDs or 10 congeners of PCDFs.

The similarity ( S ) can be used to evaluate the reliability of

he chlorine substitution modelling. The higher the similarity is,

he better the predictions agree with the measurements. On the

ther hand, the chlorine substitution probabilities determined from

he modelling can be used to examine whether the formation

f isomer-specific PCDDs and PCDFs are mainly controlled by the

hlorine substitution process. The higher the probabilities are, the

ore probably the chlorine substitution occurs.

. Results

.1. i G-CLC performance

For the PW combustion experiments in the laboratory-scale flu-

dized bed reactor, Fig. 3 (a) and (b) show the volume concentra-

ions of the main gaseous components in the first cycle. For the

eaction time, the conventional PW incineration process continues

pproximately 100 s [ Fig. 3 (a)], however the i G-CLC process of PWs

H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18 13

Fig. 3. Gas concentrations during the first cycle in the conventional incineration (a); and i G-CLC (b).

Fig. 4. The combustion efficiency under two combustion conditions.

s

f

t

c

i

i

a

a

e

t

P

d

s

C

c

s

t

t

i

r

c

i

c

c

t

u

4

i

t

P

t

P

T

c

c

T

ustains longer than 200 s [ Fig. 3 (b)]. Obviously, air is more reactive

or the combustion of PWs than the Fe-based OCs. CO 2 concentra-

ions reach the maximum value before 100 s for two combustion

onditions. The CO 2 concentration in the conventional incineration

s obviously lower than that in the i G-CLC process of PWs, which

s related to the dilution of N 2 in fluidizing gas. Meanwhile, CO is

lso generated before 100 s. This is because the pyrolysis of PWs

nd the release of volatiles occur quickly at the initial stage, how-

ver they are unable to be fully converted by either air or OC in

he laboratory-scale fluidized bed reactor. On the other hand, a few

W particles may suspend in the freeboard (due to relatively low

ensity), leading to a shorter residence time in the reactor. Fig. 3 (b)

hows that no CO 2 was detected in the oxidation stage of the i G-

LC process, indicating very high combustion efficiency and CO 2

apture efficiency in i G-CLC.

Through detecting the chloride ion concentration in NaHCO 3

aturated solution by the ion chromatograph and detecting

he absorbance of the methyl orange absorption liquid by

he UV/VIS, we have measured the amounts of HCl and Cl 2 n the CLC atmosphere: 9.7735 × 10 −4 g and 8.2824 × 10 −7 g,

espectively.

t

Fig. 4 shows the combustion efficiencies in 10 successive cy-

les for two combustion conditions. The combustion efficiency in

G-CLC is more than 99.1%. It can be accepted in the i G-CLC pro-

ess of PWs, although the efficiency is still lower than that of the

onventional incineration. Thus, it indicates that using the i G-CLC

echnology to dispose PWs is able to attain high-efficient energy

tilization as well as high-concentration CO 2 stream.

.2. Distribution of PCDD/Fs

The concentration distribution of 17 toxic PCDD/Fs congeners

n two XAD-2 resins is shown in Fig. 5 (a). There are three impor-

ant observations. ( 1 ) In i G-CLC, the concentrations of the 17 toxic

CDD/Fs congeners are obviously lower than those in the conven-

ional incineration condition. ( 2 ) For two samples, the amount of

CDFs is approximately 10 times more than that of PCDDs. ( 3 )

he OCDF represents the highest contribution to the total PCDD/Fs

ontent in the conventional incineration, while 1234678HpCDF

ontributes the highest in the i G-CLC. Fig. 5 (b) reports the

EQ distribution of 17 toxic PCDD/Fs congeners in two combus-

ion conditions. It can be found that 12378PeCDF, 123478HxCDF,

14 H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18

Fig. 5. The concentration distribution (a) and the TEQ (b) of 17 toxic PCDD/Fs congeners.

Table 1

The distribution profiles of 17 toxic PCDD/Fs congeners under dif-

ferent combustion conditions.

Combustion conditions Conventional incineration i G-CLC

PCDDs(ng/g) 13.54 0.68

PCDFs(ng/g) 101.66 5.88

PCDDs / PCDFs 0.13 0.12

Total amounts (ng/g) 115.20 6.56

Total TEQ (ng/g) 3.09 0.34

Total TEQ (ng/Nm

3 ) 0.25 0.01

Table 2

Chlorine substitutions probability and similarity under different combus-

tion conditions.

Combustion conditions Conventional incineration i G-CLC

PCDDs 1,4,6,9-substituted 0.900 0.776

similarity 0.999 0.994

PCDFs 1,9-substituted 0.955 0.848

4,6-substituted 0.831 0.589

similarity 0.983 0.946

e

s

d

i

p

6

u

m

c

t

w

a

s

c

b

i

s

a

t

d

t

t

a

c

t

23478PeCDF, 123678HxCDF, 2378TCDF and 1234678HpCDF (which

all belong to the congeners of PCDFs) are the main toxic contrib-

utors for these samples. Compared with the conventional inciner-

ation, the TEQ of 17 toxic PCDD/Fs congeners in i G-CLC is remark-

ably lowered.

The total amounts, TEQ and ratio of PCDDs/PCDFs have been

calculated for these samples. Clearly, the total amounts and TEQ

of PCDD/Fs in i G-CLC are obviously lower than from conventional

incineration. More pronouncedly, the total amounts and TEQ of

PCDD/Fs have been reduced by 94% and 89%, respectively, in the

i G-CLC in comparison of the conventional incineration condition.

Table 1 shows that the PCDDs/PCDFs ratio of the conventional in-

cineration is slightly higher than that of i G-CLC. The PCDDs/PCDFs

ratio is commonly used as a homologue pattern indicator to iden-

tify the PCDD/F characteristics [7] . It is generally believed that the

formation of PCDDs is mainly derived from the precursor conver-

sion route, while the formation of PCDFs is mainly ascribed to the

de novo synthesis route [7] . In this study, the PCDDs/PCDFs ratios

are 0.12 ∼0.13 in both conventional incineration and i G-CLC, which

indicates the de novo synthesis route dominates the dioxins forma-

tion in these experimental conditions.

4.3. Chlorine substitution model analysis

We utilized the chlorine substitution modelling to obtain the

similarity as a function of the chlorine substitution probability,

ither for PCDDs or for PCDFs. Then, the optimal chlorine sub-

titution probability with the maximum similarity was finally

etermined by gradual approximation. More details can be found

n Section S3 of SM. The predicted amounts of PCDD/Fs are com-

arable with the experimental measurements, as shown in Fig.

. Although some individual congeners are significantly over or

nder predicted, the larger yields of some important congeners

ake the impact of lower yielding congeners negligible. The un-

ertainty of the chlorine substitution modelling is originated from

he measurement error and complex formation routes (however

e assumed that higher chlorinated PCDDs or PCDFs isomers

re generated from lower chlorinated PCDDs or PCDFs isomers

tep by step). The effects of the measurement uncertainty of each

ongener to cumulative uncertainty on total toxic congeners can

e analyzed using perturbation methods.

Table 2 summarizes chlorine substitution probabilities and sim-

larities under different combustion conditions. Note that a higher

imilarity represents a less difference between model prediction

nd experimental measurement. It can be found that the similari-

ies are always higher than 98% for the formation of PCDDs during

ifferent combustion processes; for the formation of PCDFs all of

he similarities are lower, although they are above 90%. This means

hat the application of chlorine substitution model is quite reliable,

t least, for the formation of PCDDs. On the other side, a higher

hlorine substitution probability implies an easier chlorine substi-

ution for the formation of PCDDs or PCDFs, and their formation

H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18 15

Fig. 6. Simulation results vs . experimental measurements for PCDD/Fs under different combustion conditions.

i

c

1

s

o

m

s

b

d

r

d

a

c

5

5

a

T

t

t

c

t

t

t

o

d

s

t

c

r

A

Fig. 7. The Cl 2 yield under different combustion conditions.

w

y

U

t

c

f

c

g

t

I

m

w

c

s

s more likely dominated by the chlorine substitution process. For

onventional incineration, the chlorine substitution probability in

,4,6,9-position for PCDDs is 0.900 and it is 0.955 and 0.831, re-

pectively in 1,9-position and in 4,6-position for PCDFs, which is

bviously higher than those in the i G-CLC condition. These results

anifest that the chlorine substitution process is largely respon-

ible for the formation of PCDDs and has relatively less responsi-

ility for the formation of PCDFs in i G-CLC. These results also in-

icate that the chlorine substitution process plays more important

ole in conventional incineration than in i G-CLC. We therefore de-

uced that the absence of O 2 and effective dechlorination via HCl

dsorption in i G-CLC can suppress the chlorination processes to a

ertain extent and then help inhibit the formation of PCDD/Fs.

. Discussion

.1. PCDD/Fs inhibition mechanism of i G-CLC

Carbon sources, oxygen sources, chlorine sources and metal cat-

lysts are essential factors for the formation of PCDD/Fs [7–9] .

hus, we discussed the PCDD/Fs inhibition mechanism of CLC in

erms of oxygen sources and chlorine sources.

It is well known that the absence of O 2 can effectively restrict

he de novo synthesis of PCDD/Fs. Factually, the existence of O 2

an accelerate carbon residue gasification and promote the oxida-

ive decomposition of macromolecular carbon structure [7] . Among

hem, condensation, cyclization and chlorination of aromatic struc-

ures are relevant with the formation of PCDD/Fs. In i G-CLC, the

xidation breakage of macromolecular carbon structure is relatively

ifficult due to the O 2 -free reducing atmosphere.

The chlorine substitution modelling demonstrated the chlorine

ubstitution probability in i G-CLC is obviously lower than that in

he conventional incineration, which should be related to chloride

ontent (CaO can adsorb part of chlorine element) and active chlo-

ine components (the absence of O 2 leads to a lower Cl 2 yield).

ccording to the absorbance of methyl orange absorption liquids

hich were collected in the fluidized bed experiments, the Cl 2 ields in the different combustion conditions are shown in Fig. 7 .

sing the conventional incineration as a reference, one finds that

he Cl 2 yield was reduced by 82% in i G-CLC.

In the reduction stage of CaO-decorated Fe 2 O 3 /Al 2 O 3 , gaseous

hlorine source (HCl as main chloride of PW pyrolysis) can be ef-

ectively reduced via reaction (R4) 2HCl + CaO → CaCl 2 + H 2 O. The

hlorides generated (CaCl 2 ) cannot react with O 2 to reproduce

aseous chlorine source (HCl or Cl 2 ), which has been verified by

he thermodynamic calculation in our previous publication [18] .

n order to further evaluate the effect of CaO decoration, the

orphologies and surface compositions of the used OC particles

ere characterized using an Environmental Scanning Electron Mi-

roscope coupled with an Energy Dispersive X-Ray spectroscopy

ystem (ESEM-EDX, FEI Quanta 20 0 0). Indeed, chlorine can be

16 H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18

Fig. 8. PCDD/Fs inhibition and formation mechanisms in i G-CLC.

s

s

a

c

l

r

i

m

w

m

a

O

o

a

(

a

p

u

m

t

q

a

u

t

m

2

o

T

t

detected on the surface of CaO-decorated OCs after 10 redox cy-

cles, however is not detectable on the non-decorated OC surface

(see Fig. S5). This result illustrates the functionality of CaO adsor-

bent for effective dechlorination.

Hence, the PCDD/Fs inhibition mechanism of CLC can be con-

cluded from the following two aspects: on the one hand, no O 2 at-

mosphere in CLC can inhibit the de novo synthesis of PCDD/Fs; on

the other hand, a lower Cl 2 yield due to the O 2 -free atmosphere

and further dechlorination via CaO-decorated OC in i G-CLC can re-

strict precursor conversion and de novo synthesis to some extent

because active chlorine sources are one of important factors for

the chlorine substitution process through both precursor conver-

sion and de novo synthesis. Fig. 8 summarizes the PCDD/Fs inhibi-

tion mechanism of i G-CLC.

5.2. PCDD/Fs formation mechanism in i G-CLC

However, we must mention here that there are still few

PCDD/Fs formed in the O 2 -free reduction atmosphere. It is be-

lieved that active lattice oxygen (in OC) or oxygen in fuel can also

participate in the PCDD/Fs formation reactions [32] , although the

PCDD/Fs amount is far less than that under gaseous oxygen (air)

atmosphere. What is more, we notice that the i G-CLC technology

may bring disadvantageous factors for the inhibition of PCDD/Fs.

Here, we focused on the metal oxide and carbon residue, which are

requisite elements for the de novo synthesis and low-temperature

catalytic-assisted precursor conversion.

Some fine OC particles form due to abrasion and fragmen-

tation and then are brought away from FR, and are finally de-

posited in the filter. The first issue that should be addressed is,

will metal oxides (e.g . , Fe O and Al O in OCs) and chlorine ad-

2 3 2 3

orbent (e.g . , CaO) influence the formation of PCDD/Fs? A two-

tage reactor (which is composed of an air incineration reactor

nd a fixed bed reactor) was utilized to identify whether chemi-

al components in OCs catalyze or decompose the macromolecu-

ar. The PWs were first incinerated in air atmosphere in the first

eactor (the air incineration reactor, holding at 1173.15 K as a typ-

cal temperature of a refuse incinerator), where precursors, inter-

ediates and PCDD/Fs have been generated. Then the flue gases

ere introduced into the second reactor (the fixed bed reactor,

aintaining 773.15 K as a typical operation temperature for cat-

lytic reaction [33] ), which was loaded with either silica sands or

C particles (raw Fe 2 O 3 /Al 2 O 3 or CaO-decorated Fe 2 O 3 /Al 2 O 3 ). The

rganic matters in the exhaust gases of the second reactor were

bsorbed via XAD-2 resin (8 g) and toluene absorption solution

100 mL). These organic matters collected were further analyzed by

Fourier-Transform Mass Spectrometry (FT-MS, SolariX 7.0T) with

ositive ion mode for measuring the abundance of different molec-

lar weights. More details about the two-stage reactor and experi-

ental procedure can be found in Section S5 of SM.

Fig. 9 shows total ion chromatograms of the toluene absorp-

ion solutions in two cases (the fixed bed reactor was loaded with

uartz sands or CaO-decorated Fe 2 O 3 /Al 2 O 3 ) of the two-stage re-

ctor experiments. Quartz sand was used for reference. The molec-

lar weights were identified only when the abundance was more

han 2.5 × 10 7 .

As shown in Fig. 9 (a), the molecular weights with the maxi-

um intensity are 389.28 and 391.28, followed by 413.27, 274.27,

17.96, 337.17 and 484.21. It is known that the molecular weights

f 17 toxic PCDD/Fs are in the range of 306.0-460.8 [7–9] .

he result demonstrates the PCDD/Fs have been generated in

he air incineration process. What is more, the exhaust gases

H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18 17

Fig. 9. Total ion chromatograms of toluene absorption solutions in the two-stage reactor experiments.

c

F

d

1

s

a

m

a

C

o

r

o

c

T

r

e

t

e

p

d

f

c

p

(

r

l

C

p

F

C

6

d

t

f

t

m

F

b

fi

t

t

o

m

f

o

(

e

t

r

p

c

g

n

c

(

e

a

i

l

w

p

a

c

s

c

t

w

A

C

t

f

E

Z

ontain some macromolecular precursors and intermediates. From

ig. 9 (b), the molecular weight sequence according to the abun-

ance level is {389.28, 413.27, 307.26, 274.27, 300.81, 279.19, 197.08,

82.35, 323.08, 336.93}. Compared with the reference case, the

ubstances with the molecular weights of 391.28, 217.96, 337.17

nd 484.21 were not detected, while the substances with the

olecular weights of 307.26, 300.81, 279.19, 197.08, 182.35, 323.08,

nd 336.93 were generated when the flue gases passing through

aO-decorated Fe 2 O 3 /Al 2 O 3 . To be more specific, the abundance

f molecular weights 389.28 and 413.27 increases obviously. These

esults indicate that CaO-decorated Fe 2 O 3 /Al 2 O 3 could supply the

xygen needed for the formation of PCDD/Fs congeners and de-

ompose relatively small molecules (precursors and intermediates).

he effects of Fe 2 O 3 and Al 2 O 3 on the PCDD/Fs formation has been

eported [34,35] , which are consistent with our observations. Heeb

t al. [33] studied the catalytic effect of Cu element on the forma-

ion of PCDD/Fs and found that Cu element can cause the PCDD/Fs

mission to be increased by up to 3 orders of magnitude. Com-

ared with Cu element, the influences of raw Fe 2 O 3 /Al 2 O 3 or CaO-

ecorated Fe 2 O 3 /Al 2 O 3 were very small.

On the other side, the combustion efficiencies in i G-CLC are in-

erior to that in conventional incineration, which will lead to more

arbon residue in exhaust gases (escaped from the FR then de-

osited in the filter). Using an X-ray photoelectron spectroscopy

XPS, AXIS ULTRA), we have detected the carbon element and chlo-

ine element in ashes (Fig. S4 in Section S4, SM), which were col-

ected in the filter of the laboratory-scale fluidized bed system.

arbon residue, fine OC particles as well as fly ashes of PWs will

rovide requisite elements for the de novo synthesis of PCDD/Fs.

ig. 9 also demonstrates the PCDD/Fs formation mechanism of i G-

LC.

. Conclusions

The main aims of this work are to measure quantitatively the

istributions of 17 toxic congeners of PCDD/Fs in i G-CLC of plas-

ic wastes via a HRGC/HRMS and to analyse the inhibition and

ormation mechanisms of PCDD/Fs through sample characteriza-

ion and chlorine substitution model. Two combustion experi-

ents (conventional air incineration; i G-CLC using CaO-decorated

e O /Al O as OC) were conducted in a laboratory-scale fluidized

2 3 2 3

ed reactor that simulates an i G-CLC process. This work veri-

ed that the formation of PCDD/Fs can be significantly inhibited

hrough the i G-CLC technique. To be more specific, the amount and

oxic equivalency quantity of PCDD/Fs can be reduced by an order

f magnitude using the i G-CLC technique. The PCDD/Fs inhibition

echanism of i G-CLC is manifested in two aspects. First, the O 2 -

ree atmosphere in the fuel reactor of i G-CLC can suppress obvi-

usly the de novo synthesis of PCDD/Fs. Second, lower Cl 2 yield

due to the restriction of Deacon reaction in absence of O 2 ) and

ffective dechlorination (by CaO decorated to OC particles) can fur-

her inhibit the formation of PCDD/Fs through restricting the chlo-

ination processes. The chlorine substitution modelling, which re-

roduced the experimental measurements, demonstrated that the

hlorine substitution probabilities for the formation of 7 toxic con-

eners of PCDDs and 10 toxic congeners of PCDFs were indeed sig-

ificantly reduced indeed in i G-CLC.

Much smaller amounts of PCDD/Fs generated in the i G-CLC pro-

esses are partly attributed to a small amount of carbon residue

here a lower combustion efficiency was observed in the i G-CLC

xperiment, and carbon residue was detected via XPS in the fly

shes collected in the tail filter) and the presence of metal oxides

n OCs (Fe 2 O 3 , Al 2 O 3 ). The FT-MS analysis of organic matter col-

ected in the two-stage reactor experiments shows metal oxides

ill supply oxygen needed for the PCDD/Fs formation or decom-

ose the marcomoleculars, more or less, which have been gener-

ted in the incineration reactor.

To sum up, the i G-CLC of PWs attains highly efficient energy re-

overy of wastes, PCDD/Fs inhibition, and inherent CO 2 separation

imultaneously. This represents a significant advance over current

onventional waste incineration techniques and could be extended

o other chlorine contained wastes, e.g . , sewage sludge, kitchen

aste, and waste paper residue.

cknowledgments

These authors were supported by “National Key R&D Program of

hina ( 2016YFB0600801 )”, and “National Natural Science Founda-

ion of China ( 51522603 and 51561125001 )”. The authors are grate-

ul to the Analytical and Testing Center of HUST for XPS and ESEM-

DX measurements. The authors are grateful to Dioxin laboratory,

hejiang University for PCDD/Fs measurements. The authors also

18 H. Zhao, J. Wang / Combustion and Flame 191 (2018) 9–18

[

[

[

[

[

[

thank Prof. Hong Yao and Dr. Guangqian Luo in the same labora-

tory for providing access to the two-stage reactor. Prof. Fanxing Li

in North Carolina State University is also appreciated for valuable

suggestions.

Supplementary materials

Supplementary material associated with this article can be

found, in the online version, at doi:10.1016/j.combustflame.2017.12.

026 .

References

[1] H. Daniel , B.T. Perinaz , What a waste, World Bank Publishing, 2012 .

[2] H.Z. Tian , J.J. Gao , L. Lu , D. Zhao , K. Cheng , P.P. Qiu , Temporal trends and spa-tial variation characteristics of hazardous air pollutant emission inventory from

municipal solid waste incineration in China, Environ. Sci. Technol 46 (2015)10364–10371 .

[3] A. Sepulveda , M. Schluep , F.G. Renaud , M. Streicher , R. Kuehr , C. Hageluken ,A.C. Gerecke , A review of the environmental fate and effects of hazardous sub-

stances released from electrical and electronic equipments during recycling:

examples from China and India, Environ. Impact Assess. Rev. 30 (2010) 28–41 . [4] O. Eriksson , G. Finnveden , Plastic waste as a fuel – CO 2 -neutral or not? Energy

Environ. Sci. 2 (2009) 907–914 . [5] L.A. Ruth , Energy from municipal solid waste: a comparison with coal combus-

tion technology, Prog. Energy Combust. Sci. 24 (1998) 545–564 . [6] S.M. Al-Salem , P. Lettieri , J. Baeyens , The valorization of plastic solid waste

(PSW) by primary to quaternary routes: from re-use to energy and chemicals,

Prog. Energy Combust. Sci. 36 (2010) 103–129 . [7] B.R. Stanmore , The formation of dioxins in combustion systems, Combust.

Flame 136 (2004) 398–427 . [8] N.W. Tame , B.Z. Dlugogorski , E.M. Kennedy , Formation of dioxins and furans

during combustion of treated wood, Prog. Energy Combust. Sci. 33 (2007)384–408 .

[9] M. Altarawneh , B.Z. Dlugogorski , E.M. Kennedy , J.C. Mackie , Mechanisms for

formation, chlorination, dechlorination and destruction of polychlorinateddibenzo-p-dioxins and dibenzofurans (PCDD/Fs), Prog. Energy Combust. Sci. 35

(2009) 245–274 . [10] H. Zhou , A.H. Meng , Y.Q. Long , Q.H. Li , Y.G. Zhang , A review of dioxin-related

substances during municipal solid waste incineration, Waste Manag. 36 (2015)106–118 .

[11] W.Y. Lin , L.C. Wang , Y.F. Wang , H.W. Li , G.P. Chang-Chien , Removal character-

istics of PCDD/Fs by the dual bag filter system of a fly ash treatment plant, J.Hazard. Mater. 153 (2008) 1015–1022 .

[12] S.Y. Lu , T. Chen , J.H. Yan , X.D. Li , Y.L.M.J. Ni , K.F. Cen , Effects of calcium-basedsorbents on PCDD/F formation from pentachlorophenol combustion process, J.

Hazard. Mater. 147 (2007) 663–671 . [13] J.Y. Fu , X.D. Li , T. Chen , X.Q. Lin , A. Buekens , S.Y. Lu , J.H. Yan , K.F. Cen , PCDD/Fs’

suppression by sulfur–amine/ammonium compounds, Chemosphere 123 (2015)9–16 .

[14] W.T. Hsu , P.C. Hung , M.B. Chang , Catalytic destruction vs . adsorption in con-

trolling dioxin emission, Waste Manag. 46 (2015) 257–264 . [15] H. Deacon, Improvement in manufacture of chlorine, 1875, US Patent 165,802.

[16] M. Ishida , D. Zheng , T. Akehata , Evaluation of a chemical-looping-combus-tion power-generation system by graphic exergy analysis, Energy 12 (1987)

147–154 .

[17] J. Adanez , A. Abad , F. Garcia-Labiano , P. Gayan , L.F.D. Diego , Progress in chemi-cal-looping combustion and reforming technologies, Prog. Energy Combust. Sci.

38 (2012) 215–282 . [18] J.X. Wang , H.B. Zhao , Chemical looping dechlorination through adsorbent-dec-

orated Fe 2 O 3 /Al 2 O 3 oxygen carriers, Combust. Flame 162 (2015) 3503–3515 . [19] A. Lyngfelt , B. Leckner , T. Mattisson , A fluidized-bed combustion process with

inherent CO 2 separation; application of chemical-looping combustion, Chem.Eng. Sci. 56 (2001) 3101–3113 .

[20] W.Z. Bi , R.D. Zhao , T.J. Chen , J.L. Wu , J.H. Wu , Study on the formation of

PCDD/Fs in PVC chemical looping combustion, J. Fuel. Chem. Technol. 43 (2015)884–889 .

[21] R. Xiao , Q.L. Song , Characterization and kinetics of reduction of CaSO 4 with carbon monoxide for chemical-looping combustion, Combust. Flame 158

(2011) 2524–2539 . 22] J.X. Wang , H.B. Zhao , Evaluation of CaO-decorated Fe 2 O 3 /Al 2 O 3 as an oxygen

carrier for in-situ gasification chemical looping combustion of plastic wastes,

Fuel 165 (2016) 235–243 . 23] J.X. Wang , H.B. Zhao , Application of CaO-decorated iron ore for inhibiting

chlorobenzene during in situ gasification chemical looping combustion of plas-tic waste, Energy Fuel 30 (2016) 5999–6008 .

[24] A. Lyngfelt , Oxygen carriers for chemical looping combustion-40 0 0 h of oper-ational experience, Oil Gas Sci. Technol. 66 (2011) 161–172 .

25] Z.W. Xu , H.B. Zhao , Y.J. Wei , C.G. Zheng , Self-assembly template combustion

synthesis of a core-shell CuO@TiO 2 -Al 2 O 3 hierarchical structure as an oxy-gen carrier for the chemical-looping processes, Combust. Flame 162 (2015)

3030–3045 . 26] W.J. Yang , H.B. Zhao , J.C. Ma , D.F. Mei , C.G. Zheng , Copper-decorated hematite

as an oxygen carrier for in situ gasification chemical looping combustion ofcoal, Energy Fuels 28 (2014) 3970–3981 .

[27] W.J. Yang , H.B. Zhao , K. Wang , C.G. Zheng , Synergistic effects of mixtures of

iron ores and copper ores as oxygen carriers in chemical-looping combustion,Proc. Combust. Inst. 35 (2015) 2811–2818 .

[28] E. Wikstrom , S. Ryan , A. Touati , B.K. Gullett , Key parameters for de novo for-mation of polychlorinated dibenzo-p-dioxins and dibenzofurans, Environ. Sci.

Technol. 37 (2003) 1962–1970 . 29] M.H. Moon , D. Kang , H. Lim , J.E. Oh , Y.S. Chang , Continuous fractionation of fly

ash particles by SPUTT for the investigation of PCDD/Fs levels in different sizes

of insoluble particles, Energ. Environ. Sci. 36 (2002) 4 416–4 423 . [30] C. Linderholm , A. Lyngfelt , A. Cuadrat , E. Jerndal , Chemical-looping combustion

of solid fuels – Operation in a 10 kW unit with two fuels, above-bed and in--bed fuel feed and two oxygen carriers, manganese ore and ilmenite, Fuel 102

(2012) 808–822 . [31] F. Lino , T. Imagawa , B.K. Gullett , Dechlorination-controlled polychlorinated

dibenzofuran isomer patterns from municipal waste incinerators, Environ. Sci.

Technol. 34 (20 0 0) 3143–3147 . 32] Y. Qian , M.H. Zheng , W.B. Liu , X.D. Ma , B. Zhang , Influence of metal oxides on

PCDD/Fs formation from pentachlorophenol, Chemosphere 60 (2005) 951–958 .[33] N.V. Heeb , M. Zennegg , E. Gujer , P. Honegger , K. Zeyer , U. Gfeller , A. Wichser ,

M. Kohler , P. Schmid , L. Emmenegger , A. Ulrich , D. Wenger , J.L. Petermann ,J. Czerwinski , T. Mosimann , M. Kasper , A. Mayer , Secondary effects of catalytic

diesel particulate filters: copper-induced formation of PCDD/Fs, Environ. Sci.Technol. 41 (2007) 5789–5794 .

[34] H.C. Wang , S.H. Chang , P.C. Hung , J.F. Hwang , M.B. Chang , Catalytic oxidation of

gaseous PCDD/Fs with ozone over iron oxide catalysts, Chemosphere 71 (2008)388–397 .

[35] E. Voncina , T. Solmajer , Thermolysis of 2,4,6-trichlorophenol chemisorbed onaluminium oxides as example of fly ash mediated surface catalysis reaction in

PCDD/PCDF formation, Chemosphere 46 (2002) 1279–1286 .