Behavior of PCDD / Fs , PCBs , CBzs and PAHs during Thermal Treatment of Various Fly Ash from Steel Industry

The integrated iron and steel industry is the main source of polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs), or, more briefly, of dioxins. These mainly arise from the sintering process but also from other operations, such as blast furnace, basic oxygen furnace and electric arc furnace. In this study, we investigated the contribution of fly ash from the above-mentioned operations to PCDD/Fs as well as other organic pollutants (polychlorinated biphenyls [PCBs], chlorobenzenes [CBzs] and polyaromatic hydrocarbons [PAHs]). The experiments were conducted using a lab-scale vertical tube reactor, and target pollutants in off-gas and residue were collected for analysis. The experimental results show that sintering fly ash generates far greater amounts of these pollutants than other fly ash, especially in terms of PCDD/Fs and dioxin-like PCBs, which are 2–3 orders of magnitude higher. The homologue group profiles of the target pollutants also show significant differences. Both the PCBs and CBzs are dominated by low-chlorinated compounds in all samples; however, the STA sample generated a much more high-chlorinated compounds. In addition, the correlations between various target pollutants based on all samples show that PCDDs are clearly related to the low-chlorinated PCBs and CBzs, and PCDFs are strongly related with high-chlorinated PCBs and CBzs. Within each homologue group of PCDD/Fs, the isomer signature was also further scrutinized, with special emphasis on the seventeen 2,3,7,8-substituted PCDD/Fs, as well as the seven PCDD-congeners and two TCDF usually associated with chlorophenol precursor routes, with the goal of shedding more light on the mechanism of PCDD/Fs-formation.


INTRODUCTION
Many hazardous pollutants are emitted from iron and steel production, including polychlorinated dibenzo-pdioxins (PCDDs) and dibenzofurans (PCDFs), or briefly "dioxin".On the basis of the inventory of dioxins emissions to the Chinese environment in 2004, dioxin discharged by steel and other metal smelting industries reached 45.6% of total dioxin emission load of 10.24 kg year -1 , which was the largest contribution.In the European Union, iron ore sintering has overtaken municipal solid waste incineration (MSWI) as the largest source of PCDD/Fs emissions (Tuppurainen et al., 1998).With the rise in the amount of crude steel production in China, nearly attaining 803.8 million tons in 2015, the more dioxin emission and the considerable harm to human health and environment are happening (Kao et al., 2011;Nie et al., 2012).
Integrated steelworks are characterized by networks of interdependent material and energy flows between the various production units, most of which are sintering plants, pelletisation plants, coke oven plants, blast furnaces, basic oxygen steel-making plants and electric arc furnace with subsequent casting.The sintering plant preparing iron ore fines as feed for the blast furnace was known as the main source of PCDD/Fs (Buekens et al., 2001).The sintering process transforms the moist agglomerated raw material (a mixture of fine iron ores, coke breeze (3-5 wt.%), limestone, recycled reverts, fluxes and other materials) into strong sintered and partly-reduced ore particles.The process is finished when the flame front reaches the bottom of the bed.
Relative to the position of the flame front, the sintering bed is subdivided into four reaction and heat exchange zones, from top to bottom: air preheating in the sintered zone, coke ignition and sintering zone (flame front), pre-heating zone, and wet quenching zone (Ooi and Lu, 2011).Presumably, most PCDD/Fs is generated in the pre-heating zone, few cm in front of the flame zone, where temperature rapidly drops from > 650 to < 200°C.PCDD/Fs-formation in the sintering bed has been attributed to two factors: (1) transition metal salts and chlorides volatilize in the flame zone and transfer with combustion gas, subsequently condense, and react with (2) volatilized organic compounds, coke or soot in a temperature and oxidation window, favorable for PCDD/Fs-formation (from 500°C down to 200°C) (Nakano et al., 2005;Nakano et al., 2009;Ooi and Lu, 2011).Therefore, the zone for PCDD/Fs-formation is still driven by the flame front and the cooling effect of the dried raw materials, yet bounded by the quenching effect exerted by the moist layer at the bottom.As this layer grows thinner, various pollutants are no longer absorbed in the moist bed and gradually break through.This observation forms the basis of the recirculation of final and dirty off-gas to the first zones of the belt.According to these views, the bulk of the PCDD/Fs output originates in the bulk of the burden, representing by far the largest part of the iron ore material.For PCDD/Fs emissions control of flue gas, many technologies have been developed, such as adsorption by carbon materials (Li et al., 2016), inhibition by suppressors (Lin et al., 2015) and catalytic destruction (Yu et al., 2016).
Studies about the PCDD/Fs formation on the whole process of steel making are rather less abundant, which are mainly focus on iron ore sintering (Buekens et al., 2001;Kasai et al., 2001;Suzuki et al., 2004;Nakano et al., 2005Nakano et al., , 2009;;Ooi and Lu, 2011;Zhang and Buekens, 2016).For sintering, it is generally considered that PCDD/Fs are formed through the de novo synthesis because PCDD/Fs generated from this process show the typical homologue group profiles developed from this pathway.Kasai et al. (2001) found that the raw materials on de novo synthesis generated very low PCDD/Fs, whereas adding sintering fly ash increased dioxins formation markedly.During the European Union (E.U.) project "Minimization of dioxins in thermal industrial processes: mechanisms, monitoring and abatement" (MINIDIP), the evolution of PCDD/Fs, PCBs, CBzs, CPs and PAHs from fly ash samples changing strongly with the experimental conditions (reaction temperature and time, oxygen and moisture content, addition of various inhibitors and activated charcoal) was systematically investigated (Buekens et al., 2001;Zhang and Buekens, 2016).Many studies have focused on the influence of operating conditions (temperature, sintering time, excess oxygen and bed permeability) and raw materials (metal catalysts, chlorine and solid fuel) on PCDD/Fs formation in iron ore sintering.However, there is the current lack of PCDD/Fs emission levels of other steelmaking processes and the connection between PCDD/Fs isomers and indicator compounds in steel industry.In addition, various fly ashes generated from different steelmaking processes are usually recycled as sintering raw material, because of high content of Fe.
The formation mechanisms of dioxins is still derived from the research on MSWI and model/real fly ash (Huang and Buekens, 1995;Stieglitz et al., 1997;McKay, 2002;Stanmore, 2004;Yang et al., 2004;Cunliffe and Williams, 2007;Altarawneh et al., 2009;Fujimori et al., 2009;Wang et al., 2010;Chiu et al., 2011).Two main formation pathways have been proposed: de novo synthesis and precursor pathway.De novo synthesis refers to formation of PCDD/Fs from a catalytic partial oxidation of carbon or soot, which is also characterized by the PCDDs/PCDFs-ratio greater than unity and a decreasing concentration order of CBzs > CPs > PCDFs > PCDDs > dioxin-like PCBs (polychlorinated biphenyls) (Huang and Buekens 1995;Ooi and Lu, 2011).Precursor synthesis is supported by fast condensation of chlorophenols (CP) to PCDDs and PCDFs congeners, or, following a longer route, starting from chlorobenzenes (CBzs) or other suitable structures such as Polycyclic Aromatic Hydrocarbons (PAHs).Both theories have been linked because the de novo route also generates precursors largely, including CPs and CBzs.A large number of research papers and reviews focus on examining the relationship between PCDD/Fs and other organic pollutants, such as CPs, CBzs and PAHs, for guidance to study the formation mechanisms of dioxin further (Wikström et al., 1999;Blumenstock et al., 2000;Tuppurainen et al., 2000;Ryu et al., 2004).Wikström et al. (1999) found that the PCDDs are mainly generated by chlorphenols condensation, while the PCDFs are formed through a non-or a low-chlorinated precursor followed by further chlorination reactions.Besides CPs as an important precursor for PCDD/Fs formation, Iino et al. (1999) reported that PCDFs and polychlorinated naphthalene (PCN) could also be formed directly from PAHs.Despite many studies on PCDD/Fs formation mechanisms, there is no clear route to explain this problem.
The aim with this study was to experimentally investigate the formation of PCDD/Fs, PCBs, CBzs and PAHs from five different fly ash which were sampled from various steelmaking processes in an integrated iron and steel works including sintering, blast furnace, basic oxygen furnace and electric arc furnace.More importantly, the entire isomer signature was scrutinized, with special emphasis on those congeners associated with 2,3,7,8-substituted PCDD/Fs and chlorophenol precursor routes.Meanwhile, this study had investigated the interrelationships among PCDD/Fs isomers and multiple precursors, such as PCBs, CBzs and PAHs.With these results, we hope to throw more light on the mechanism of PCDD/Fs-formation in thermal metallurgical processes.

Experimental Materials
In this study, all five fly ash samples were collected from different process in an integrated iron and steel plant including sintering, blast furnace, basic oxygen furnace and electric arc furnace.The sintering fly ash (STA), basic oxygen furnace fly ash (BOA) and electric arc furnace fly ash (EAA) were sampled from electrostatic precipitator.The blast furnace fly ash was collected from gravity settler and Venturi-tube, which was defined as BFA1 and BFA2, respectively.All fly ash samples were dried for 24 h in an oven at 105°C and sieved under 100 mesh.

Experimental Set-up and Procedure
Organic pollutant (PCDD/Fs, PCBs, PAHs and CBzs) formation tests were conducted in a vertical tube furnace system, as shown in Fig. 1.The apparatus comprises a quartz glass tubular reactor (height 60 cm, diameter 2 cm), a temperature controller and a thermocouple for monitoring the reaction temperature of the sample bed.In each test, 1 g fly ash sample was positioned in the middle of quartz reactor tube, after the furnace temperature reached the temperature set value.A synthetic gas stream of 100 mL min -1 passed through the sample, accompanied by the generation of targeted pollutants.Each fly ash sample was treated at 350°C with 12% oxygen for 1 h, and the residence time was 1.9 s.
The target pollutants evolving in the gas phase were absorbed on XAD-2 resin and then by toluene in consecutive gas scrubbing bottle.After the test, the cooled test residues were collected together with the absorbed volatiles for further analysis.All tests for PCDD/Fs and PCBs formation were obtained in duplicate to ensure the accuracy.The arithmetic average of the duplicated results was used in our further analysis.The variability of PCDDs, PCDFs and PCBs defined as a percentage of the standard deviation to the average value is in a range from 0.9-16.4%,except for PCDDs of BFA2 (28.8%) and PCDFs of BFA1 (20.5%).

Analytical Methods
The clean-up procedure of the PCDD/Fs and PCBs samples followed the US EPA method 1613 (U.S. EPA, 1999) and EPA method 1668 (U.S. EPA, 2008), respectively.A HRGC/HRMS analysis separated, identified and quantified the various congeners of PCDD/Fs and PCBs using a 6890 Series gas chromatograph (Agilent, USA) coupled to a JMS-800D mass spectrometer (JEOL, Japan).A DB-5ms (60 m length × 0.25 mm ID, 0.25 µm film thickness) capillary column separated 38 PCDDs and 56 PCDFs individual congeners or combinations out of 136 PCDD/F congeners (P = 4 to 8).Similarly, All 209 PCB isomers from monochlorinated (MoCB) to decachlorinated biphenyls (DeCB) were detected, and over 160 congeners were separated and quantified.The temperature program of GC oven was optimized as follows: (1) For PCDD/Fs: splitless injection of 1 µL, initial oven temperature of 150°C maintained for 1 min, then raised at a rate of 20 °C min -1 to 190°C, and finally increased at 3 °C min -1 to 280°C as final temperature and held for 20 min.(2) For PCBs: splitless injection of 1 µL, initial oven temperature of 90°C maintained for 1 min, then raised at a rate of 20 °C min -1 to 180°C, held for 1 min and finally increased at 3 °C min -1 to 300°C as final temperature and held for 20 min.
Pretreatment of CBz samples were according to the state method of HJ/741-2015 (China, 2015).Quantitative analysis of CBzs concentrations proceeded by gas chromatography using an Agilent 6890N GC, equipped with electron capture detector (ECD) and a 30 m DB-5ms column.Oven temperature initially was maintained for 4 min at 80°C, then raised to 106°C at 5 °C min -1 and to 250°C at 8 °C min -1 , the final temperature was held for 4 min.The PAH samples were analyzed using a GC/MS (JMS-Q1050, Japan) machine equipped with a DB-5ms column after the clean-up procedure on the basis of HJ 646-2013646- (China, 2013) ) and EPA method 8270 (U.S. EPA, 1998).The temperature program for GC oven was as follows: splitless injection of 1 µL, initial temperature of 50°C held for 2 min, then increased at 15 °C min -1 to 200°C and held for 5 min, finally increased at 10 °C min -1 to 300°C and held for 5 min.
The toxic equivalents (TEQ) were calculated by the North Atlantic Treaty Organization (NATO) factors, in agreement with the industrial practice in the E.U. and China.Meanwhile, the contribution of PCBs to WHO-TEQ values usual in Japan was also established, which typically remains marginal, at few %.

RESULTS AND DISCUSSION
Temperature is a major factor of influence in dioxins formation (McKay, 2002;Zhang and Buekens, 2016).In this study, the temperature for various organic pollutants formation is 350°C, which was identified as the optimal temperature for PCDD/Fs-formation on fly ash (Zhang and Buekens, 2016).Because various fly ashes collected from different electrostatic precipitator are used as the recycled material in the sintering, the 12% oxygen in this study match with the real oxygen concentration which ranges from 5% to 15% in an iron ore sintering bed (Kasai et al., 2001).

The Yield of Various Organic Pollutants
The amount of PCDDs, PCDFs, PCBs and dl-PCBs generated from five fly ashes are presented in Fig. 2(a).
The yield of organic pollutants generated from EAA, BFA1, BFA2 and BOA sample are almost at the same level, which ranges from 1.8-2.4ng g -1 , 1.8-2.1 ng g -1 and 822.7-934.9ng g -1 for PCDDs, PCDFs and PCBs, respectively.The total amount of PCDDs and PCDFs in STA sample, with a value of 258.6 ng g -1 and 1360.2 ng g -1 respectively, are 2 to 3 orders of magnitude higher than the value of other four fly ash samples.However, PCBs concentration of STA sample is 1620.7 ng g -1 , which is only twice as much as other samples.The TEQ of all samples ranges from 67.3-77.3pg WHO-TEQ g -1 except STA sample, which is much higher, nearly attaining 50,673.5 pg WHO-TEQ g -1 .The ratio of PCDFs/PCDDs is similar for almost all samples in a range of 1.0-1.7,except for STA sample with a value of 5.3.The PCDFs/PCDDs ratio > 1 suggests that PCDD/Fs form mainly through de novo synthesis, especially for sintering fly ash.Except for STA sample which is 6.4, the average chlorination degree (Cl-PCDD/Fs) (Lin et al., 2015) for all samples is between 5.2 and 5.6 (Table 1).The abundance of both KCl and NaCl may provide a suitable condition for de novo synthesis of PCDD/Fs in the STA sample (Kuo et al., 2011).Buekens et al. (2001) found the same decreasing order of CBzs > PCBs > PCDFs > PCDDs when the ESP samples were annealed at 350°C.However, when the heating temperature was 250, 300 and 400°C, the descending order of different organic pollutants was CBzs > PCDFs > PCBs > PCDDs.Obviously, 350°C is the optimal temperature for PCDD/Fs and PCB formation, but the rate of PCBs increase is much faster than PCDD/Fs with the temperature elevated.Thus, the temperature at 350°C of this research is the main reason that the PCBs concentration is higher than that of PCDD/Fs.The yield of PAHs and CBzs in the thermal treatment of five fly ashes are shown in Fig. 2(b).The results show that the yield of PAHs and CBzs in STA sample with a value of 3.1 µg g -1 and 125.0 µg g -1 respectively, are much higher than other four kinds of fly ash samples.The emission level of CBzs and PAHs from these four samples are in a range of 0.8-1.1 µg g -1 and 52.1-84.0µg g -1 , respectively.
The concentration of organic pollutants emitted from all samples are in decreasing order of PAHs >> CBzs > PCBs > PCDFs > PCDDs > dl-PCBs, with small shifts in individual sample, such as dl-PCBs > PCDDs in EAA and BFA1.Obviously, compared with other four fly ashes, fly ash from sintering plant generates much more PCDD/Fs and dl-PCBs, by more than 2-3 orders of magnitude.However, the amount of other pollutants including PCBs, PAHs and CBzs generated from STA sample, is only two to four times than that of others.
Table 2 presents the elemental analysis of five fly ash samples.The chlorine content of STA sample is 24.8%, however, the chlorine content of EAA sample is the highest among other four fly ash samples, with only 1.1%.Cieplik et al. (2003) reported that the PCDD/Fs output doubled with 8 ppm (wt) of chloride added as NaCl, but with the same proportion of chlorine administered as C 2 Cl 4 , the dioxin output was over 2 orders of magnitude larger.Suzuki et al. (2004) found that the dioxin concentration of the sinter mixture under conditions of zero HCl supply was only one-sixth of that with 100 ppm HCl supply.Meanwhile, the average degree of chlorination in the case with no HCl supply was lower than that with HCl supply in both PCDDs and PCDFs.The results of elemental analysis of five fly ash samples indicate that higher PCDD/Fs generation and higher Cl-PCDD/Fs of sintering fly ash should be ascribed to higher chloride content than other four fly ash samples.BFA1 and BFA2 fly ash samples are similar, both for the yield and distribution of all organic pollutants, as well as the elements.It suggests that the different fly ash from the same source have the similar dioxin formation pathway, which is unrelated with initial particle size.

Distribution of Various Organic Pollutants
Fig. 3(a) shows signatures of the ten homologue groups, from TCDD through OCDF (basis: PCDDs + PCDFs = 100%).Sample BFA1 and BFA2 collected from the same blast furnace present the similar homologue profiles.It suggests that particle size have little influence on PCDD/Fs formation.The signatures of sample EAA and BOA are similar, in which the percentage of low-chlorinated PCDFs are much more than high-chlorinated PDCFs.In the abovementioned samples, PCDDs are largely dominated by TCDD and OCDD, however, STA sample presents a totally different signature.In the STA sample, PCDDs are largely dominated by OCDD, and PCDFs are occupied by highchlorinated PCDFs.
Principal component analysis (PCA) is used to illustrate the similarities and differences between congener distributions and de novo formation.The matrix is used for statistically investigating the homologue profiles of PCDD/Fs in all fly ash samples (Fig. 4).In this study, no transformation of data is performed during statistical analysis.Two main clusters (A and B) are presented in the loading diagram separated by factor 1 and factor 2. Factor 1 and factor 2 explains 74.7% and 20.2% of the total variance, respectively.Cluster A only contains sample STA and cluster B includes all other fly ash samples.It suggests that sintering fly ash leads to a distinctly different pathway for PCDD/Fs formation.
Signatures of the ten homologue groups, from MoCB through DeCB are shown in Fig. 3(b).Except for STA   DiCBz and HxCBz dominate the CBzs emission in sample STA, whereas DiCBz and TrCBz account for the majority of the CBzs emission in other four fly ash samples, as shown in Fig. 3(c).In sample STA, the emission proportion of high-chlorinated CBzs are much more than that of other samples, which also reaches an agreement with stronger chlorination of sintering fly ash.The signatures of three homologue groups of PAHs are shown in Fig. 2(d), indicating almost no difference for all tested samples.14 PAH compounds are combined into "Other PAHs" group because of low and parallel proportion of them accounting for all 16 PAH compounds concentration in all five samples.The main compound of these 16 PAHs is naphthalene (Nap) accounting for more than 90%, followed by the acenaphthylene (Acy).Fly ash from sintering plant generates much less Acy than other fly ash samples, there is only one-tenth left.

Relationships between PCDD/Fs and Other Compounds
For each target compound, the concentration is similar in EAA, BFA1, BFA2 and BOA sample, however, it is considerably different in STA sample.Thus, five data sets are only equivalent to two sets including STA data set and other four samples data sets.This fact leads to the result that the relationships between PCDD/Fs and other compounds built by the actual concentration are inaccurate and meaningless.For example, all correlation coefficients (R) calculated by actual concentration of various compounds are nearly 1.00, it is invalid to determine the relationship among these compounds.Thus, the data sets used in this section are the relative weight (%) of homologues in their respective compounds.
In the PCA score plot (Fig. 5), 38 variables are performed separately including PCDD/Fs (10), PCBs (10), CBzs (4) homologues and all PAH compounds (14) except indene(1,2,3-cd)pyrene, dibenzo(ah)anthracene and benzo(g,h,i)perylene, which are not detected in all samples.Two cluster (A and B) are clearly separated by factor 1 and factor 2 in the loading diagram.Obviously, cluster A includes all PCDF homologues (except for TCDF) as well as the all high-chlorinated PCBs (penta-, hexa-, hepta-, octa-, nona-and deca-) and all high-chlorinated CBzs (penta-and hexa-).Oppositely, cluster B located at the right of the ordinate axis is comprised of all PCDD homologues (except.for HpCDD), all low-chlorinated PCBs (mono-, di-, triand tetra-) and all low-chlorinated CBzs except for DiCBz The result shows that PCDD-and PCDF-congeners in this study are related to low-and high-chlorinated target compounds (PCBs and CBzs) respectively, indicating that the similar formation mechanisms of various compoundsin the same cluster.However, all PAH compounds are dispersed over a wide area, it suggests that the formation The correlation coefficient is applied to estimate the correlation between PCDD/Fs, PCBs and CBzs (Table 3).Because the data used in this section are the relative weight (%) of homologues in their respective compounds, the correlation coefficients of PCDDs are actually the opposite of PCDFs for all homologues.The selectivity of PCDDs are strongly linked (R) with low-chlorinated PCBs (mono-, di-, tri-and tetra-) and CBzs (tri-and tetra-).Conversely, PCDFs are clearly antagonistic to all of that, which are strongly related with high-chlorinated PCBs and CBzs.This result is consistent with consequence of principal component analysis in Fig. 5.It suggests that PCDDs may generate from low-chlorinated PCBs and CBzs, and PCDFs may form by reactions of high-chlorinated PCBs and CBzs.Meanwhile, the correlation between the PCDD/Fs and CBzs concentration is investigated because CBzs are the most important indicators for dioxins online monitoring.The results show an excellent correlation (R = 0.99) between PCDD/Fs and high-chlorinated CBzs, including PeCBz and HxCBz, it suggests that PeCBz and HxCBz show ideal potential as indicators for PCDD/Fs.This result is consistent with previous findings (Yan et al., 2012).

Isomer Distribution Pattern of PCDD/Fs
In our earlier work, isomer distribution patterns had been already investigated systematically.This study is a part of a much more comprehensive study, confronting PCDD/Fs-yields and signatures both from combustion tests and de novo tests, and covering a variety of distinct iron and steel plant conditions: sintering mix, sintering fly ash and other process fly ash (the present study), as well as sludge from continuous casting, hot rolling and cold rolling.Some of these test series featured the addition of suppressants including spent anion exchange resins, calcium oxide and slaked lime.

CONCLUSIONS
In this study, the emission levels and signatures of PCDD/Fs, PCBs, CBzs and PAHs from five fly ash samples collected from different processes of the iron and steel industry were systematically investigated during a set of de novo tests.
The yields of the target pollutants (PCDD/Fs, PCBs, CBzs and PAHs) generated from the EAA, BFA1, BFA2 and BOA samples are almost at the same level; however, the STA sample generated far more of the above-mentioned pollutants than the other samples.It is noteworthy that fly ash from the sintering plant generated much greater amounts of PCDD/Fs and dl-PCBs, by more than 2-3 orders of magnitude.
The homologue group profiles of the target pollutants also show significant differences between the STA sample and the other samples.Both the PCBs and CBzs are dominated by low-chlorinated compounds in all samples; however, the STA sample generates many more highchlorinated compounds than the other samples, suggesting that the STA sample has much stronger chlorination for PCBs and CBzs formation.In addition, the correlations between PCDD/Fs, PCBs and CBzs are evident.PCDDs are clearly related to the low-chlorinated PCBs (mono-, di-, tri-and tetra-) and CBzs (tri-and tetra-), and PCDFs are strongly related with high-chlorinated PCBs (penta-, hexa-, hepta-, octa-, nona-and deca-) and CBzs (penta-and hexa-).
Meanwhile, a scrutiny of fingerprints are obtained at all the samples.This study focused on the chlorophenol routes and 2,3,7,8-substituted PCDD/Fs in investigating the formation mechanisms of PCDD/F congeners in the steel manufacture process.

Table 1 .
The PCDFs/PCDDs-ratio and Cl-PCDD/Fs-values of all samples.

Table 2 .
The elemental analysis of five fly ash samples (wt.%).