**Removal of Cu (II) from Industrial Wastewater Using Mechanically Activated Serpentinite**

**Petros Petrounias 1,\*, Aikaterini Rogkala 1, Panagiota P. Giannakopoulou 1, Paraskevi Lampropoulou 1, Petros Koutsovitis 1, Nikolaos Koukouzas 2, Nikolaos Laskaris 3, Panagiotis Pomonis <sup>4</sup> and Konstantin Hatzipanagiotou <sup>1</sup>**


Received: 19 February 2020; Accepted: 23 April 2020; Published: 3 May 2020

**Abstract:** We investigate with this study the effectiveness of mechanically activated serpentinite in capturing Cu (II) from the multi-constituent acidic wastewater of the pit lakes of the Agios Philippos mine (Greece), proposing specific areas with serpentinites suitable for such environmental applications. For this purpose ultramafic rock samples that are characterized by variable degrees of serpentinization from ophiolitic outcrops exposed in the regions of Veria-Naousa and Edessa have been examined regarding their capacity to remove the toxic load of Cu (II) from wastewater after having been mechanically activated through a Los Angeles (LA) machine (500, 1000 and 1500 revolutions). The more serpentinized and mechanically activated samples, as they have been characterized after a combination of various mineralogical, petrographic, geochemical analyses as well as after different stresses of abrasion and attrition, seem to be more effective in Cu removal than the less serpentinized ones. Selective removal of Cu (II) in the wroewolfeite phase was obtained by using the mechanically activated highly serpentinized ultramafic rocks. Furthermore, areas with highly serpentinized ultramafic rocks defined after petrographic mapping, using GIS method, which can potentially be used as filters for the effective Cu (II) removal from industrial wastewater are suggested.

**Keywords:** removal of Cu; mechanical activation of serpentinite; sustainability; wastewater treatment

#### **1. Introduction**

Today, societies are vastly affected by negative changes in the quality of water, air, and soil as a result of human activities. More specifically, large volumes of wastewater present within lakes, river or in industries, put intense pressure on mankind. Among the different types of pollutants, non-organic ones have a high-risk factor and are characterized as being harmful [1]. Mitigation of toxicity factors identified within water considered as waste that has been produced by industrial activities is a main concern in modern-day societies. Wastewater produced from industrial and human activities, must be brought back the societies as well as to Nature, which includes rivers, lakes and land [2]. More specifically, acidic drainage or pit lakes produced by past mining activities are held as being responsible for numerous problems associated with degraded water quality of which is responsible for

limited access to water, especially for people living in surrounding regions with high poverty rates, raising risk factors for both conflicts and environmental reduction potentials for natural regeneration processes to occur [3–6]. Pit lakes are often acidic (pH 2–3), containing elevated concentrations of metals and metalloids showing a potential for the formation of highly acidic conditions, due to the fact that they exhibit relatively low buffer capacity levels [7–9]. Heavy metals are considered as high risk pollutants which are present in wastewater, including those originated from mining operations and industries which produce plants, batteries, paints and pigments, which also include industries that produce glass and ceramic materials. Wastewater often incorporates elements such as Cr, Pb, Cd, Zn, Cu and Ni [9]. When these metals are exposed to the environmental conditions, it is observed that ions are accumulated within humans, a process that will take place by consumption in the whole food chain cycle, thereby creating toxic concentrations of metals characterized as heavy that must not reach the environment [10].

Several applications have been proposed to achieve the removal of the toxic load, which include coagulation, chemical precipitation, filtration, ion exchange and solvent extraction, as well as enablement of evaporation and membrane methodologies [11]. Adsorption of heavy metals can be realized with conventional materials such as activated carbon which is often used effectively in various applications and also the use of carbon formed by organic materials (carbonized). Low cost carbon-containing materials have been used for the adsorption of inorganic pollutant materials and the removal of heavy metals, such as agricultural wastes (wool, rice, rice hulls, peat moss, cork biomass, untreated coffee grounds), modified biopolymers and industrial by-products [12,13]. Recent studies [14,15] state that peat can be regarded as highly effective in removing several heavy metals (like Hg) derived from water solutions by the performance of two types of experiments (batch and column). Peat produced by sphagnum plant moss accumulation is very effective at absorbing and removing Cd hypochlorite and oxidized Cd(CN)2 plant waste, Cr6<sup>+</sup> derived from aqueous solutions, and also Pb, Ni, Cu, Zn and Cd from waters considered as waste [16–19]. Biochar can be formed from various sources that include, for example: straw, shells, crop residues, wood, stover, bark, rubber, sludge, litter and peat [20–23]. Biochars produced from peat moss using pyrolysis procedures with variable carbonization conditions have raised research interest for heavy metal adsorption [24].

Numerous researchers have mentioned the use of many different rock lithotypes that include dolomite, magnesite, limestone, andesite, serpentinite and dunite not only in order to increase the pH value of industrial wastewater but also to capture heavy metals such as Cu [25,26]. Most of the aforementioned rock types are broadly used combined with other rocks as aggregate materials in a large number of construction and industrial applications while producing a vast number of sterile materials [27]. Therefore, it is important to investigate methods in order to dispose of these sterile aggregates, in accordance with the principals of sustainability, recycling, alternative use of materials, by concurrently reducing the necessity of additional energy for mineral and rock extraction. The sustainable use of natural rocks in removing heavy metals constitutes a Nature-based self-purification function, focusing on achieving "zero wastes" and "zero emissions" goals.

Serpentinite has been used as an adsorbent of Cu from industrial wastewater [26] despite the fact that the risk exists for the leaching of Cr and Ni. Its abundance in several ophiolite complexes, its wide range of construction and industrial applications combined with its particular textural characteristics make it particularly attractive for such applications.

It is widely accepted by the broad scientific community that different hydrated silicate minerals and especially clay minerals (that encompass serpentine), demonstrate abilities in the adsorption of undesirable heavy metal elements. Serpentine constitutes a group of high-Mg 1:1 layered trioctahedral minerals, with its structural formula being Mg6(Si4O10)(OH)8. It is characterized as chemically simple but from the view of texture can be present in various forms. The serpentine mineral group occurs in three polymorphs: the high pressure antigorite and the low pressure chrysotile and lizardite. Lizardite in particular displays an ideal layered topology and on the other hand chrysotile has a bent structure and antigorite is often modulated [28,29]. Serpentine polymorphs belong to the phyllosilicate mineral group that displays a 1:1 layered structure, which comprises tetrahedral and octahedral sheets [28]. Four distinct O-H chain groups were identified in serpentine minerals. These are classified into two distinct topologies based on their (OH) positioning. Three of the four O-H groups are situated on the inner surface amongst two layers, whereas the third is located within the layer. The primitive unit serpentinite cell is held responsible for the cohesive attachment of the two successive layers that form hydrogen bonds [28,30].

When serpentinite is attached through the application of mechanical force activation (e.g., ball impact), this results in structural changes that mostly affect the OH groups that could potentially be brought to a looser state in the octahedral positions, enhancing the dispersal of the latter by the effects of water. When the serpentine structure is damaged through grinding, it becomes incapable of maintaining atom arrangements in a proper order to preserve its crystallinity. Scientists like Huang [31] have used activated serpentine in order to remove Cu from wastewater. The raw serpentine without grinding operation adsorbs less Cu than the chemo-mechanically-activated serpentine.

The aim of this study is the selective capture of Cu (II) within the crystalline structure of serpentine from tmechanically activated serpentinized ultramafic rocks from the acidic wastewater of a pit lake of the Agios Philippos mine (Greece). The suggestion of specific areas in Greece that present serpentinized ultramafic rocks suitable for Cu capture is a supplementary goal of this study.

#### **2. Geological Description of Rock Materials Sources**

Representative serpentinized rocks deriving from the Veria-Naousa and Edessa ophiolite complex (Greece) have been selected for studying their effectiveness in the removal of Cu from industrial wastewater after having been mechanically activated. The Veria-Naousa ophiolite belongs to the Almopias subzone (Axios geotectonic zone) in northern Greece. Its sequence consists of intense and highly tectonized serpentinized lherzolite and harzburgite of 46.76 km2 total surface area (calculated via ArcMap 10.1), penetrated by scarce pyroxenitic dykes [32], and from gabbro, diabase and pillow basalt (Figure 1).

Remnants of oceanic lithosphere (Upper Jurassic to Late Cretaceous) constitute the Edessa ophiolite complex which was thrust out of one or more ocean basins [34–36]. This complex presents intense tectonization and consists of several tectonic units [37,38] such as serpentinized ultramafic rocks (lherzolite and harzburgite) of 50.34 km<sup>2</sup> total surface area (calculated via ArcMap 10.1). Moreover, diabase is the main mafic lithotype of the complex, whereas gabbro, basalt as well as diorite are less frequent (Figure 2).

**Figure 1.** Modified geological map of Veria-Naousa ophiolite complex [33] after fieldwork and mapping using ArcMap 10.1; the area of investigation is shown in the black rectangle; the green color shows the serpentinized ultramafic rocks of the region.

**Figure 2.** Modified geological map of Edessa ophiolite complex [39] after fieldwork and mapping using ArcMap 10.1; the area of investigation is shown in the black rectangle; the green color shows the serpentinized ultramafic rocks of the region.

#### **3. Materials and Methods**

Taking into account how effective the serpentinite was in removing Cu from the wastewater from the pit lake from the Agios Phillippos Kirkis mine [26], four serpentinites, derived from the aforementioned ophiolite complexes and presenting different degree of serpentinization were used after their mechanical activation to various degrees to examine the mechanism through which serpentinite can become more effective in the removal of Cu.

#### *3.1. Characterization of the Studied Rock Materials*

The first stage for the characterization of the investigated rocks is their petrographic examination (mineralogical and textural features) with the aid of a Leitz polarizing optical microscope (Leica Microsystems, Wetzlar, Germany). The modal composition of the investigated serpentinites was also identified using powder X-ray diffraction (XRPD) analysis, with a Bruker D8 Advance Diffractometer (Bruker, Billerica, MA, USA), with Ni-filtered CuKα radiation. The scanning area for bulk mineralogy of the samples covered the 2θ interval 2–70◦, with a scanning angle step size of 0.015◦ and a time step of 0.1 s. The modal composition was identified with the DIFFRACplus EVA 12® software (Bruker-AXS) based on the ICDD Powder Diffraction File of PDF-2 2006. The mineral phases were semi-quantified with the TOPAS 3.0 ® software (TOPAS MC Inc., Oakland, CA, USA), based on the Rietveld method refinement routine. The routine is based on the calculation of a single mineral-phase pattern regarding to the crystalline structure of each mineral, and the refinement of the pattern using a non-linear least squares routine. According to the study by Bish and Post [40], the quantification errors were calculated and they are estimated to be ~1%. Polished thin sections of the examined rock samples were analyzed in scanning electron microscopes (SEM) in order to identify their mineralogical characteristics. Serpentine minerals microanalyses were carried out in a JEOL JSM-6300 SEM equipped with energy dispersive (EDS) using the INCA software (version). The scanning electron microscope used is located in the Laboratory of Electron Microscopy and Microanalysis (University of Patras, Greece). Operating conditions were accelerating voltage 25 kV and beam current 3.3 nA, with a 4 μm beam diameter. The total counting time was 60 s and dead-time 40%. Synthetic oxides and natural minerals were used as standards for the analyses, where the detection limits are ~0.1% and accuracy better than 5% was obtained. Furthermore, an XRF (X-Ray fluorescence) spectrometer and a sequential spectrometer (ICP-ES) were used for the determination of the major and trace elements of the studied samples, which were carried out at Bureau Veritas Mineral Laboratories (Vancouver, BC, Canada).

#### *3.2. Methodology for the Cu (II) Removal*

The studied serpentinites were placed in the Los Angeles (LA) machine in order to become mechanically activated after they have been sieved in the No. 14 sieve (1.40 mm). The Los Angeles (LA) machine is a rotating drum which contains certain number of steel spheres. Los Angeles (LA) test constitutes a basic test which indicates the mechanical quality of aggregates rocks by identifying their resistance in abrasion and attrition. At this stage, serpentinite samples were rotated in the LA machine for 500, 1000 and 1500 revolutions, respectively. Then, the samples were sieved appropriately in order to reach the grain size of 0.8–0.6 mm which then used as filters in batch type columns. Serpentinites that were not mechanically activated were also used as filters in the batch type columns after having been sieved appropriately in order to reach a grain size between 0.8 and 0.6 mm.

The next stage included an experimental arrangement consisting of columns of borosilicate glass of 20 mm diameter. In each column 70 gr of the different mechanically activated serpentinites (derived from the different revolutions used in the LA machine) was used, and 300 mL of wastewater was passed through each column four times. After this four pass procedure, wastewater samples were further processed for geochemical analysis of Cu concentration, which was performed in the Institute for Solid fuels Technology and Application of the National Centre for Research and Technology Hellas using conjugated plasma argon mass spectrometry. All the sorption experiments were carried out at room temperature (25 ± 2 ◦C).

#### **4. Results**

#### *4.1. Results of the Studied Rock Materials*

#### 4.1.1. Petrographic Features of the Studied Rock Materials

Two types of serpentinized ultramafic rocks regarding their degree of serpentinization were collected from the studied areas. Their textural and mineralogical characteristics are analyzed below in Figure 3.

**Figure 3.** Textural features of serpentinites from the investigated ophiolite complexes: (**a**) Photomicrograph of mesh texture in serpentinized (srp) matrix (sample BE.01, + Nicols); (**b**) Back-scattered electron image (BSE) of Cr-spinel (Cr-sp) with curved boundaries and thin rims of ferritchromit and magnetite as well as mesh serpentine (srp) (sample ED.111); (**c**) Photomicrograph of an orthopyroxene (opx) porphyroclast and clinopyroxene porphyroclast (cpx) surrounded by olivine neoblasts (ol) in serpentinized (srp) matrix (sample BE.118, + Nicols); (**d**) Photomicrograph of orthopyroxene porphyroclast (opx) and olivine grains (ol), subhedral Cr-spinel (Cr-sp) and ribbon serpentine (srp) (sample ED.86B, + Nicols).

#### (1) Highly-serpentinized (Group I)

The primary modal mineralogical assemblage of the highly serpentinized ultramafic rocks is intensely altered, whereas relics of orthopyroxene and Cr-spinel were observed. The subhedral to euhedral Cr-spinel are frequently crosscut by secondary veins and/or surrounded by thin rims of magnetite and ferritchromit because of the effects of ocean floor alteration processes. Serpentine is the main alteration phase which forms mesh, ribbon, bastite and hourglass textures. Moreover, less chlorite and magnetite subsequently observe, likely being the result of retrograde metamorphism during exhumation (Figure 3).

#### (2) Medium-Serpentinized (Group II)

The primary assemblage of these serpentinized ultramafic rocks comprises of orthopyroxene, olivine, less of clinopyroxene, olivine and Cr-spinel constituting less than 30% of the whole assemblage. Porphyroclasts of orthopyroxene exhibit exsolution lamellae of clinopyroxene, typical characteristic of mantle peridotites. Olivine displays porphyroclastic grains and smaller neoblasts. Locally, olivine porphyroclasts show strain lamellae, undulose extinction, shearing and recrystallization (Figure 3). Clinopyroxene appears as relict subhedral porphyroclasts. Crystals of Cr-spinel presented as subhedral to euhedral and they display an irregular distribution of ferritchromit compositional areas. The boundary between the Cr-spinel (either unaltered or altered) and the ferritchromit is curved and lobate. The main secondary product is serpentine which predominantly displays ribbon and mesh textures, whereas others are chlorite and magnetite. The modal composition of the investigated serpentinites was further determined by XRPD analysis in Figure 4.

**Figure 4.** XRPD patterns of the examined serpentinites: (**a**) highly serpentinized ultramafic rock (sample: BE.01); (**b**) highly serpentinized ultramafic rock (sample: ED.111B); (**c**) medium serpentinized ultramafic rock (sample: BE.118); (**d**) medium serpentinized ultramafic rock (sample: ED.86B), (1: Serpentine, 2: Spinel, 3: Magnetite, 4: Clinopyroxene, 5: Orthopyroxene, 6: Olivine).

#### 4.1.2. XRPD of Mineral Rock Materials

The X-ray diffraction enabled us to identify the crystalline phases of the tested ultramafic rocks where higher picks of serpentine were observed in samples BE.01 and ED.111B whereas samples BE.118 and ED.86B display lower picks of serpentine (Figure 4). Serpentine content as well the content of other mineralogical phases of the studied rock samples were calculated via Rietveld method and are listed in the Table 1, where the sample BE.01 presents as the most serpentinized, while ED.86B presents as the less serpentinized sample. The identification of the degree of serpentinization of the studied rock samples through the petrographic analysis via polarizing microscope is in accordance with the results of Rietveld method.

**Table 1.** Semi-quantitative mineralogical assemblage of the tested serpentinites. The quantification errors calculated for each phase according to Bish and Post [40] are estimated to be ~1% (ol: olivine, opx: orthopyroxene, cpx: clinopyroxene, sp: spinel, mgt: magnetite, srp: serpentine, BE: Veria-Naousa ophiolite, ED: Edessa ophiolite) (-: below detection limit).


#### 4.1.3. Chemistry of Serpentine Minerals

Representative microanalyses of the serpentine minerals are shown in Table 2 and plotted in the diagrams of Figure 5. Serpentine minerals from Veria-Naousa and Edessa ophiolite are composed of SiO2 (42.02–46.27 wt.%), MgO (36.45–41.50 wt.%), Fe2O3 (1.10–6.53 wt.%) and less Al2O3, CaO and TiO2. Serpentine minerals from Group I display higher SiO2, MgO, Fe2O3 and lower Al2O3, CaO and TiO2 contents than those of Group II. NiO and Cr2O3 contents are wide (0.00–2.20 wt.% and 0.00–1.93 wt.%, respectively). This wide range of NiO and Cr2O3 content may be connected with the nature of the replaced olivine in the case of Ni-rich serpentine and pyroxene as well as in the case of Cr-rich serpentine.

**Figure 5.** (**a**) MgO vs. SiO2 plot and (**b**) FeO vs. MgO plot for the analyzed serpentine minerals in serpentinites from Veria-Naousa and Edessa ophiolites. Fields of lizardite, chrysotile and antigorite are from Singh & Singh [41].


#### *Energies* **2020** , *13*, 2228

In the diagram of Figure 5a, the analyzed serpentine minerals from Group I are mostly antigorite, while Group II are mostly lizardite and chrysotile and less antigorite. The similarity in characteristics is consistent with the plot of MgO vs. FeO (Figure 5b). FeO was calculated by the Fe2O3 method in order produce the diagram shown in Figure 5b.

#### 4.1.4. Geochemical Features of Rock Materials

Major and trace elements data from the studied serpentinites, along with their loss on ignition (LOI) values are listed in Table 3. LOI values vary significantly from 7.9% to 14.6% with the most serpentinized sample (BE.01) presenting the higher LOI value, while the least serpentinized one (ED.86B) presenting the lowest LOI value.


**Table 3.** Representative geochemical analyses of the studied serpentinites (-: below detection limit, total).

#### *4.2. Experimental Study Results*

4.2.1. Chemical Analysis of the Wastewater

The wastewater from the selected pit lake of the Agios Philippos mine is characterized by an acidic pH value (2.99) and was geochemically analyzed by Petrounias et al. [26]. These results are given in Table 4. According to the results of Table 4, the wastewater contains an extremely high concentration of Cu (8847.21 ppb).


**Table 4.** Chemical analysis of wastewater (-: below detection limit) [26].

4.2.2. Chemical Analysis of Wastewater after Having Penetrated 4 Times through Columns of Batch Type

Table 5 displays the final results of the chemical analyses of wastewater after having been passed four times through each column, in which the different mechanically activated serpentinites were used as filters. As displayed in Table 5, high amounts of Cu have been trapped from the samples that contain higher serpentinite contents.


**Table 5.** Chemical analysis of the wastewater.

More specifically, the highly serpentinized sample (BE.01) seems to present a higher Cu removal capacity in comparison to the other three rock samples, even when used without having been mechanically activated, as well as when it was mechanically activated after 500, 1000 and 1500 revolutions, respectively.

4.2.3. Chemical X-ray Diffractometry of Serpentinites after the Experimental Study

The mechanically activated serpentinites (after 1500 revolutions), after having been used as filters in the experimental arrangement of the batch type columns for the removal of Cu, were analyzed by X-ray diffractometry. The corresponding X-ray diffraction patterns are given in Figure 6. In the pattern of samples BE.01 and of ED.111B, which are the most serpentinized, the mineral wroewolfeite [Cu4(SO4)(OH)62H2O] appears, in contrast to the other two samples which are characterized by lower serpentinite contents (samples BE.118 and ED.86B).

**Figure 6.** XRPD patterns of the investigated serpentinites: (**a**) highly serpentinized ultramafic rock (sample: BE.01); (**b**) highly serpentinized ultramafic rock (sample: ED.111B); (**c**) medium serpentinized ultramafic rock (sample: BE.118); (**d**) medium serpentinized ultramafic rock (sample: ED.86B), (1: Serpentine, 2: Spinel, 3: Magnetite, 4: Clinopyroxene, 5: Orthopyroxene, 6: Olivine, 7: Wroewolfeite).

#### **5. Discussion**

Several researchers have studied the individual use of minerals and of rocks materials in various environmental applications and more specifically for the removal of toxic loads using these materials [25,26]. Furthermore, quite many researchers have carried out mechanical, chemical and mechanochemical activation of minerals and rocks in order to increase their effectiveness in removing toxic loads from wastewater. Huang et al. [31] have concluded that mechanochemically activated serpentine presents very satisfactory results concerning Cu (II) removal (to an almost complete degree), allowing this type of removal to act as a way/method of recycling and reusing Cu derived from various wastewater solutions. Mechanical activation constitutes a way of applying mechanical stress to induce changes in the surface properties as well as in the crystalline structure of minerals [42–45]. Even though numerous studies relative to the understanding of the nature of the effect of mechanical loads on structural changes in the structure of minerals have been carried out, the activation of low cost phyllosilicate minerals, such as serpentine, has not yet been extensively investigated in environmental metal neutralization uses.

Petrounias et al. [26] combined sterile raw materials by using sterile aggregates of the LA test, and achieved significant removal of Cu from the Agios Philippos Kirkis mine (Greece). They attributed the Cu removal to the existence of mechanically activated serpentinite. This was carried out under pH 4 conditions, where precipitation of metal complexes is not favored [46]. In general, the pH of the solution controls the adsorption of Cu in the laminated serpentine, mainly of tetrahedral silicate during the ion exchange process.

This study, which is based on and expands the initial study of Petrounias et al. [26] attempted to find out, more accurately, the mechanism as well as the necessary and sufficient conditions for a serpentinized ultramafic rock to work efficiently in selective Cu capture from wastewater. As presented in Table 5, it is obvious that as the mechanical stress increases (through the LA test), which means that as the loadings of abrasion and attrition increase, the capacity of the serpentinized rocks in capture Cu increases regardless their degree of serpentinization. Thus, this increase may happen exclusively due to the existence and the structure of serpentine. When serpentine is subjected to mechanical loading there is structural change, especially in the OH− group, which may become more relaxed than in the octahedral positions and easily circulate in solution when in contact with it [31]. When the structure of serpentine is decomposed by the use of abrasive mechanical stresses, is is unable to retain its atom structure and thus its crystallinity is altered. This change in the crystallinity of serpentine has been cited as one of the major factors for copper adsorption [26,31]. It is understood that mechanical stress may rupture the tetrahedral and octahedral sheets of serpentine. Moreover, intensively significant index relative to the capacity of Cu capture seems to be the degree of serpentinization of the rock materials used. More specifically, the more serpentinized ultramafic rocks (BE.01, ED.111B), as they have been identified through the petrographic observation (Figure 3), the X-Ray diffractometry and Rietveld method (Figure 4, Table 1) and the geochemical analysis given through the LOI index (Table 3), seem more effective in Cu capture (Table 5) relative to those contained less serpentine (BE.118, ED.86B). Moreover, the effectiveness, in the Cu capture, of the most serpentinized ultramafic rocks is indicated by the presence of wroewolfeite in the structure of these rocks after their use as filters (Figure 6).

The amount of serpentine found in the studied rocks is presented as the determinant factor, as through this, the rate of change of available crystalline meshes capable of Cu capture from the studied solutions is determined. However, the combination of the abovementioned factors seems to constitute the more crucial combination concerning the effectiveness of each serpentinized lithotypes for the removal of Cu. More specifically, the highly serpentinized ultramafic rocks (BE.01, ED.111B with 88.6–91.7% of serpentine contained) present faster and more efficient capture of Cu in contrast to those characterized by lesser amounts of contained serpentine (BE.118, ED.86B with 59.0–70.0% of serpentine).

Furthermore, the Cu capture capacity seems to be related to the type of serpentine and therefore to the special texture characteristics of the studied serpentinites. More specifically, the serpentinized samples of similar degree of serpentinization (Table 1) but with a variety in textural features (mesh, ribbon) and in serpentine type (chrysotile, antigorite and lizardite), as shown in the corresponding diagrams (Figure 5), display significant differences regarding their Cu capture effectiveness. Serpentine from highly serpentinized ultramafic rocks (Group I) plotted in the field of antigorite (Figure 5) present as more capable of removing Cu in contrast those of Group II (medium serpentinized ultramafic rocks) whose serpentine is plotted in the field of lizardite and chrysotile. This may happen due to the different capacity of lizardite, chrysotile and antigorite for retaining their crystallinity under mechanical stress. Additionally, the petrographic characteristics such as texture may significantly influence the Cu capture. More specifically, the mesh texture of serpentine in the most serpentinized sample (BE.01) is presented as the most effective texture than the other textures of the rest studied rocks, a fact which may happen due to surface adsorption of Cu2<sup>+</sup> when CuOH<sup>−</sup> is simultaneously generated and adsorbed within the microcells.

#### *Proposed Areas with Serpentinites for Their Potential Use as Filters for Cu Removal*

Nowadays, in the era of cyclical economy and climate change, the ability to recover and reuse metals has been studied by many researchers [31,47]. Such research is fundamental to modern cutting-edge studies. In this study, the amount of Cu captured using serpentinites from the Veria-Naousa and Edessa ophiolite complexes suggest promising results to potentially recover copper accumulated within the crystallinity of serpentine (Figure 6), by using a variety of physicochemical recovery methods. The main object of the this work is to capture Cu within the crystallinity of serpentine from serpentinized ultramafic rocks and also to examine this application in conjunction with the recovery of

copper from wastewater, as well as to propose potential areas that encompass serpentinized ultramafic rocks suitable for these applications (Figure 7).

**Figure 7.** Modified geological map of Veria-Naousa ophiolite complex [33] after fieldwork and mapping using ArcMap 10.1; the area of investigation is shown in the black rectangle; the dark green color shows the highly-serpentinized ultramafic rocks of the region and the light green color shows the medium-serpentinized ultramafic rocks.

GIS-based petrographic mapping of serpentinites feasible to be used as raw materials for copper capture are suggested. In the modified maps (see below) we have taken into account all of the aforementioned factors and propose specific areas that are defined and measured from the examined ophiolite complexes (Figures 7 and 8). Specifically, the areas marked with dark green color (zone A) constitute the suggested areas from the mentioned complexes with the suitable serpentinites. These serpentinites can potentially be used as filters for Cu removal regarding their petrographic characteristics (degree of serpentinization, textural features and serpentine type) and geochemical characteristics (LOI index). The Veria-Naousa ophiolite complex comprises of highly-serpentinized ultramafic rocks, and GIS-based calculations yield an area of 43.84 km2, whereas the Edessa ophiolite complex is calculated to be 42.19 km<sup>2</sup> (Figure 7). On the other hand, zone B, marked with light green color, constitutes the lesser suitable areas for Cu capture, as they contain lower amounts of serpentine and different mineralogical and petrographic characteristics than those of zone A.

**Figure 8.** Modified geological map of Edessa ophiolite complex [39] after fieldwork and mapping using ArcMap 10.1; the area of investigation is shown in the black rectangle; the dark green color shows the highly-serpentinized ultramafic rocks of the region and the light green color shows the medium-serpentinized ultramafic rocks.

Concerning the Veria-Naousa ophiolite complex, 2.93 km<sup>2</sup> of highly serpentinized ultramafic rocks was calculated using GIS method, whereas in the Edessa ophiolite complex 8.15 km2 was calculated (Figure 8). Conclusively, the highly serpentinized ultramafic rocks from both ophiolite complexes are present in higher volumes than those of the medium-serpentinized rocks. Additionally, through the proposed maps, which are first introduced through this study, an extra use for these serpentinites is proposed apart from aggregates for construction and environmental applications as has already highlighted by Petrounias et al. [48,49].

#### **6. Conclusions**

In this paper, the selective capture of Cu (II) from acidic wastewater derived from the pit lakes of the Agios Philippos mine (Greece) from mechanically activated serpentinite using a LA machine was studied. The abovementioned study leads to the following conclusions:


Furthermore, areas with highly serpentinized ultramafic rocks that can potentially be used as filters for the effective Cu (II) removal from industrial wastewater are suggested and more specifically these are the 43.84 km<sup>2</sup> areas from the Veria-Naousa ophiolite and the 42.19 km2 area from the Edessa ophiolite.

**Author Contributions:** P.P. (Petros Petrounias) participated in the fieldwork, the elaboration of laboratory tests, the interpretation of the results, coordinated the research and the writing of the manuscript; A.R. participated in the fieldwork, performed the SEM work, the interpretation of the results and contributed to the manuscript writing; P.P.G. participated in the elaboration of laboratory tests, the interpretation of the results and contributed to the manuscript writing; P.L. carried out the XRPD analyses, participated in the interpretation of the results and contributed to the manuscript writing; P.K. participated in the elaboration of laboratory tests and in the interpretation of the results; N.K. contributed to the manuscript writing; N.L. participated in the fieldwork and in the interpretation of the results; P.P. (Panagiotis Pomonis) contributed to the manuscript writing; K.H. participated in the interpretation of the results. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We would like to thank A.K. Seferlis of the Laboratory of Electron Microscopy and Microanalysis, University of Patras for his aid, E. Gianni for her assistance in the preparation of the studied samples for laboratory tests. We also thank M. Kalpogiannaki for her assistance in the construction of the geological maps.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **NO***x* **Emissions and Nitrogen Fate at High Temperatures in Staged Combustion**

#### **Song Wu 1,\*, Defu Che 2, Zhiguo Wang <sup>1</sup> and Xiaohui Su <sup>1</sup>**


Received: 4 June 2020; Accepted: 8 July 2020; Published: 10 July 2020

**Abstract:** Staged combustion is an effective technology to control NO*x* emissions for coal-fired boilers. In this paper, the characteristics of NO*<sup>x</sup>* emissions under a high temperature and strong reducing atmosphere conditions in staged air and O2/CO2 combustion were investigated by CHEMKIN. A methane flame doped with ammonia and hydrogen cyanide in a tandem-type tube furnace was simulated to detect the effects of combustion temperature and stoichiometric ratio on NO*<sup>x</sup>* emissions. Mechanism analysis was performed to identify the elementary steps for NO*x* formation and reduction at high temperatures. The results indicate that in both air and O2/CO2 staged combustion, the conversion ratios of fuel-N to NO*x* at the main combustion zone exit increase as the stoichiometric ratio rises, and they are slightly affected by the combustion temperature. The conversion ratios at the burnout zone exit decrease with the increasing stoichiometric ratio at low temperatures, and they are much higher than those at the main combustion zone exit. A lot of nitrogen compounds remain in the exhaust of the main combustion zone and are oxidized to NO*<sup>x</sup>* after the injection of a secondary gas. Staged combustion can lower NO*<sup>x</sup>* emissions remarkably, especially under a high temperature (≥1600 ◦C) and strong reducing atmosphere (SR ≤ 0.8) conditions. Increasing the combustion temperature under strong reducing atmosphere conditions can raise the H atom concentration and change the radical pool composition and size, which facilitate the reduction of NO to N2. Ultimately, the increased OH/H ratio in staged O2/CO2 combustion offsets part of the reducibility, resulting in the final NO*x* emissions being higher than those in air combustion under the same conditions.

**Keywords:** chemical simulation; NO*<sup>x</sup>* emission; staged combustion; high temperature; strong reducing atmosphere

#### **1. Introduction**

Nitrogen oxides (NO*x*) are one of the most predominant pollutants in coal-fired boilers. Because NO*x* can endanger human health severely and cause acid rain, there is an increasing public demand for reducing NO*x* emissions. Many countries have promulgated new NO*x* emission limits [1–4]. In China, the allowed NO*<sup>x</sup>* emissions should be below 100 mg of NO2/Nm<sup>3</sup> (6% O2) for all coal-fired power plants after 2014. In the European Union, the NO*x* emission limit is expected to be lowered to 200 mg of NO2/Nm3 (6% O2) for power plants over 500 MWe by the year of 2016.

To achieve these stringent NO*<sup>x</sup>* emission limits, a combination of two or more NO*<sup>x</sup>* reduction techniques has to be used [5,6]. Currently, commercially available NO*x* reduction techniques include air staging, reburning, low-nitrogen burner [7], selective catalytic reduction (SCR) and selective non-catalytic reduction (SNCR). Among them, air staging, reburning and low-nitrogen burner are low NO*x* combustion techniques, whereas SCR and SNCR are post-combustion NO*x* reduction technologies

employed to offer varying degrees of NO*<sup>x</sup>* control capability. Because of high capital and operating costs, ammonia leakage, deactivation of the catalyst, and relatively narrow temperature windows for post-combustion NO*x*-reduction technologies [8–13], it is more desirable in practice to remove as much NO*x* as possible during combustion to alleviate the dependency on SCR and SNCR.

Air staging is the most widely used NO*x* control technique during combustion in pulverized coal boilers, which employs a main combustion zone and a burnout zone within the furnace region (see Figure 1) [5,14,15]. About 80% of air flow enters the main combustion zone where pulverized coal burns under substoichiometric conditions, i.e., reducing atmosphere conditions. The balance of the required air for complete combustion is introduced to the burnout zone through overfire air (OFA) ports. In the main combustion zone, the deficiency of oxygen can accelerate CH*m* (with *m* = 1, 2, and 3) and NH*<sup>n</sup>* (with *n* = 0, 1, and 2) to reduce generated NO*<sup>x</sup>* to N2 [16]. After the balance of oxygen is supplied to the burnout zone, a small amount of fuel NO*<sup>x</sup>* will be formed here, as most of the fuel-N has been converted into N2 or NO*x*. Meanwhile, the generation of thermal NO*<sup>x</sup>* is also inhibited. As a result, lower emissions of NO*x* are achieved at the furnace exit.

**Figure 1.** Pulverized coal boiler air-staged combustion NO*x* control system.

Many research efforts have been devoted to the characteristics of NO*<sup>x</sup>* formation and reduction in staged air combustion [1,3,16–22]. The results revealed that NO*<sup>x</sup>* emissions are mostly determined by coal properties, combustion temperature, stoichiometric ratio (SR) and residence time. Some investigators [17,22] reported that stoichiometric ratio can strongly affect the generation of NO*x*, and the conversion ratio of fuel-N to NO*<sup>x</sup>* falls with the decreasing stoichiometric ratio under reducing atmosphere conditions (SR < 1). Taniguchi et al. [1,19] detected the impact of combustion temperature on NO*x* emissions in a drop-tube furnace system used to simulate the air-staged combustion characteristics of actual pulverized coal boilers. They agreed that raising the temperature of the main combustion zone under reducing atmosphere conditions helps the reduction of NO*x*. This viewpoint is also supported by some other research work [21]. Thus, it is found that a high temperature or strong reducing atmosphere in the main combustion zone can lower NO*x* emissions. In other words, to achieve the minimal NO*x* emissions in staged air combustion, the combustion temperature should be raised as high as possible and the stoichiometric ratio should be reduced as far as possible simultaneously in the main combustion zone. Bai et al. [16] examined NO*<sup>x</sup>* emission levels of various coals under high

temperature and strong reducing atmosphere conditions in a vertical tandem-type drop-tube furnace system. Their results verified the NO*x* removal potential of this combustion method.

Creating high temperature and strong reducing atmosphere conditions in staged air combustion to lower NO*<sup>x</sup>* emissions has a promising prospect. The slag-tap furnace is expected to be the most appropriate application for the combustion method owing to the following two aspects: The first is the high combustion temperature. For example, the gas temperature within the cyclone barrel is more than 1650 ◦C in cyclone-fired boilers [23]. The second is the decreased ash melting temperature under strong reducing atmosphere conditions [24]. The slag-tap furnace prefers to fire the coals having low ash melting temperatures without severe slagging. Moreover, there are also some other furnaces featuring high combustion temperatures [5,21].

O2/CO2 combustion is one of the most promising CO2-capture technologies in thermal power generation [25–28]. Studies have shown that staged combustion can also lower the NO*x* emissions in O2/CO2 combustion [16,20,26,29]. Thus, NO*<sup>x</sup>* formation and reduction under high temperatures and strong reducing atmosphere conditions in staged O2/CO2 combustion are also worthy being explored.

A number of researchers [4,6,20,30–35] have investigated the processes of NO*<sup>x</sup>* formation and reduction in staged air and O2/CO2 combustion to date. However, these studies were performed under the conditions of relatively low temperature or relatively high stoichiometric ratio. There are very few studies on the reaction mechanisms for NO*x* under high temperatures and strong reducing atmosphere conditions, which differ from those under conventional conditions of staged combustion. Meanwhile, how to co-ordinate the combustion temperature and stoichiometric ratio to achieve the most suitable conditions for NO*<sup>x</sup>* reduction is still unclear and insufficiently understood. Therefore, it is of great significance to study how the high temperature and strong reducing atmosphere conditions influence the NO*x* formation and reduction in staged combustion.

In the present study, the characteristics of NO*x* formation and reduction under high temperature and strong reducing atmosphere conditions in staged air combustion were investigated numerically by CHEMKIN. A methane flame doped with ammonia and hydrogen cyanide for fuel-N in a tandem-type tube furnace was simulated to probe the effects of combustion temperature and stoichiometric ratio on the NO*x* emissions. Based on the calculations, the elementary steps for NO*x* formation and reduction at the high temperature were identified. In addition, NO*<sup>x</sup>* formation and reduction in staged O2/CO2 combustion were also examined.

#### **2. Numerical Approach**

#### *2.1. Reactors and Models*

The simulations were carried out for a tandem-type tube furnace consisting of two identical tube reactors. The tube furnace is shown schematically in Figure 2a. The inside diameter and heating length of the tube reactor are 38 mm and 600 mm, respectively. The heaters are arranged around the tube reactors to control the reaction temperatures. The primary gas, which includes CH4, NH3, HCN and air (O2/CO2), is introduced into the tube reactor 1 where methane burns under different temperatures and stoichiometric conditions. Meanwhile, NH3 and HCN are converted into NO or N2. Subsequently, the secondary gas, i.e., supplementary air or O2/CO2, is injected as OFA from the connection between the tube reactor 1 and 2. The residual fuel burns out in the tube reactor 2. Therefore, the tube furnace can be employed to describe the characteristics of NO*x* formation and reduction in staged combustion with the tube reactor 1 and 2 regarded as the main combustion zone and the burnout zone, respectively.

Indeed, some key processes in real boilers such as strong turbulence, devolatilization, char combustion and thermal radiation are simplified for the purpose of seeking elementary steps for N conversion at high temperatures. During pulverized coal combustion, considerable fractions of C and N conversion occur in the gas phase. When pulverized coal is ignited in the furnace, the volatiles in coal are first released and then mixed with air for homogeneous combustion. Hydrocarbons are the important components of the volatiles. Compared with the real combustion process of

pulverized coal, the transformation of Char-N and heterogeneous reduction of NO*x* are not considered. This simplification certainly brings deviation for direct prediction of NO*x* emissions in practice, and its influence depends on the amount of the coal volatiles. In this study, although methane combustion simulation cannot fully reflect the N conversion during coal combustion, it can still reveal NO*<sup>x</sup>* formation and reduction in the homogeneous combustion.

**Figure 2.** Schematic diagram of simulation object: (**a**) tube furnace of staged combustion; (**b**) reaction process modeling, 1—external source of inlet gas i, 2—plug flow reactor i, 3—external source of inlet gas ii, 4—non-reactive gas mixer, 5—plug flow reactor ii, 6—outlet flow of reactor.

The simulations were performed using a chemical kinetics modeling code CHEMKIN. It provides a feasible and powerful tool to understand reaction processes involving elementary gas-phase chemical kinetics [30–32,34,36–38]. Proper models were chosen to simulate the tube furnace and a corresponding reaction process diagram was developed in Figure 2b. Two external source of inlet gas models were used to introduce the primary and secondary gases into the reaction system. Moreover, two plug flow reactor (PFR) models were employed to describe the combustion processes in the tube reactor 1 and 2, respectively. The PFR model assumes that no mixing occurs in the flow direction while perfect mixing occurs in the direction perpendicular to the flow [39,40]. Many researchers have applied it to simulate the complex physical and chemical phenomena in tube reactors. A non-reactive gas mixer model was used to replace the connection between the two tube reactors. The reaction system ended up with an outlet flow of reactor model. All the models above completely constituted the simulation object.

#### *2.2. Reaction Mechanism*

There are three mechanisms responsible for the NO*<sup>x</sup>* formation in combustion systems: thermal NO*x*, prompt NO*x* and fuel NO*x*. In the present study, the production of NO is far more than those of other nitrogen oxides, thus, only NO is taken into account in our results. The thermal NO*x* is formed by the direct oxidation of nitrogen from the combustion air at a very high temperature (>1800 K). This reaction process can be expressed by the extended Zeldovich mechanism as follows [41]:

$$\text{N}\_2\text{+O} \rightarrow \text{NO} + \text{N}\tag{1}$$

$$\text{N} + \text{O}\_2 \rightarrow \text{NO} + \text{O} \tag{2}$$

$$\text{N} + \text{OH} \rightarrow \text{NO} + \text{H} \tag{3}$$

The prompt NO*x* are generated by the reaction of atmospheric nitrogen with hydrocarbon radicals in fuel-rich conditions. The key reactions are written by [42]:

$$\text{N}\_2\text{+CH}\rightarrow\text{HCN}+\text{N}\tag{4}$$

$$\text{NH}\_2 + \text{CH}\_2 \rightarrow \text{HCN} + \text{NH} \tag{5}$$

Subsequently, these resultants are oxidized to NO quickly. The fuel NO*<sup>x</sup>* are produced by the oxidation of nitrogen bound in the fuel and generally account for more than 80% of the total NO*<sup>x</sup>* production in large pulverized coal boilers [41]. NH3 and HCN are the dominant intermediates during the conversion of fuel-N to NO or N2. As a result, the overall reactions of the fuel NO*<sup>x</sup>* formation can be given by:

$$\text{NH}\_3 + \text{O}\_2 \rightarrow \text{NO} + \dots \tag{6}$$

$$\text{HCN} + \text{O}\_2 \rightarrow \text{NO} + \dots \tag{7}$$

In addition, the generated NO is also reduced to N2 simultaneously, which mainly depends on the local environment. The final NO*<sup>x</sup>* emissions are the comprehensive result of these actions, and a detailed reaction mechanism is needed to predict it.

A GRI-Mech 3.0 reaction mechanism was adopted in this paper, which involves 53 species and 325 elementary chemical reactions [43]. The purpose of this mechanism is to model natural gas combustion, including NO formation and reduction and reburn chemistry. The three NO*x* formation mechanisms above are all included in this mechanism. Species concentrations in reaction systems are calculated from the net rate of production for each species by chemical reaction. Reaction rate constants are determined by the modified Arrhenius expression [32]:

$$k = AT^{\emptyset} \exp(-E/[RT])\tag{8}$$

where *A* is the pre-exponential factor, *T* is the reaction temperature, β is the correction factor, *E* is the activation energy and *R* is the molar gas constant. The reverse reaction rate constants derive from the forward reaction rate constants and appropriate equilibrium constants. Under this mechanism, the rate of production (ROP) and first-order sensitivity analyses were used to interpret the kinetic results [32,34,36,44]. The ROP analysis can provide the information of the rates of formation and consumption for each species involved in the mechanism. The first-order sensitivity analysis is able to obtain the first-order sensitivity coefficient defined as:

$$\kappa = \frac{\delta Y\_j / Y\_j}{\delta A\_i / A\_i} \tag{9}$$

where *Yj* is the mole fraction for the *j*th species and *Ai* is the pre-exponential factor for the *i*th reaction. The coefficient reflects the relative change in the predicted concentration for *j*th species caused by increasing the reaction rate constant for the *i*th reaction.

#### **3. Data Analysis and Simulation Conditions**

The stoichiometric ratio (SR) is often used to express combustion conditions, which is represented in this study by:

$$\text{SR} = \frac{V\_{\text{O}\_2} / V\_{\text{CH}\_4}}{\left(V\_{\text{O}\_2} / V\_{\text{CH}\_4}\right)\_{\text{st}}} \tag{10}$$

where *V* is the volume flow rate and the subscript st denotes the stoichiometric condition. The conversion ratio of fuel-N to NO*x* (NO*x* CR) is defined as:

$$\text{NO}\_{\text{x}}\text{CR} = \frac{\text{Exhaust NO}\_{\text{x}} \text{ volume flow rate}}{\text{Inflow fuel} - \text{N volume flow rate}} \tag{11}$$

where the Inflow fuel-N volume flow rate is the sum of NH3 and HCN volume flow rates.

Nine different stoichiometric ratios in the main combustion zone (0.5–1.2, 2) were used to study the effect of the SR on the characteristics of NO*<sup>x</sup>* emissions, as shown in Table 1. The C/N mole ratios in these cases were all selected as 85. The reaction temperature in the main combustion zone varies from 1200 to 1800 ◦C, while that in the burnout zone varies from 1100 to 1400 ◦C. To compare with air-staged combustion, NO*<sup>x</sup>* emissions in staged O2/CO2 and O2/Ar combustion were also investigated with the O2 concentrations set at the same value (21%).


**Table 1.** Simulation conditions for staged combustion in the tandem-type tube furnace.

#### **4. Results and Discussion**

#### *4.1. Model Validation*

In order to obtain creditable and reasonable simulation results, a comparison between different reaction mechanisms and models was carried out in Figure 3. An updated reaction mechanism of Glarborg et al. and a premixed flame model (PFM) were also taken into consideration in the present study. The updated Glarborg reaction mechanism includes 97 species and 779 elementary chemical reactions, and is able to predict the experimental results correctly [32]. The PFM can compute species and temperature profiles in steady-state burner-stabilized premixed laminar flames. Figure 3 gives the predicted results of three cases: PFR model and GRI-Mech 3.0 reaction mechanism, PFR model and updated Glarborg reaction mechanism, PFM model and GRI-Mech 3.0 reaction mechanism [45]. The calculation was conducted under the SR of 0.7 and the reaction temperature varying from 1200 to 1800 ◦C; the other conditions are listed in Table 1. On the whole, the predicted volume flow rates of NH3, HCN and NO at the main combustion zone exit show similar trends in the three cases. Although there are some differences between the reaction mechanisms and models, these results are comparable. Especially at high temperatures, a good agreement is observed. Therefore, the above-described numerical approach is valid.

#### *4.2. Characteristics of NOx Emissions in Staged Air Combustion*

Figure 4 shows the NO*<sup>x</sup>* CRs at the main combustion zone and burnout zone exits in the staged air combustion. The NO*<sup>x</sup>* CRs at the main combustion zone exit increase when the SR rises, and the reaction temperature seems to have little influence on the NO*<sup>x</sup>* CRs. However, the NO*<sup>x</sup>* CRs at the burnout zone exit are significantly affected by the SR and reaction temperature in the main combustion zone. When the reaction temperature is low, the NO*x* CRs decrease with the increasing SR. Comparing the NO*<sup>x</sup>* CRs of the main combustion zone exit and the burnout zone exit, a large quantity of NO*<sup>x</sup>* is produced after the injection of secondary gas, which means a lot of nitrogen compounds exist in the exhaust of the main combustion zone. This point will be proved and discussed in Figure 5. The smaller the SR is, the higher the number of nitrogen compounds. Contrastingly, for the high reaction temperature, the final NO*<sup>x</sup>* emission levels are quite low. A minimal difference is found in the NO*<sup>x</sup>* CRs between the main combustion zone exit and the burnout zone exit, which means most

of the fuel-N has been converted into NO*<sup>x</sup>* or N2 and few nitrogen compounds remain at the main combustion zone exit. In addition, the lower NO*<sup>x</sup>* emission levels are found at a high temperature (≥1600 ◦C) and strong reducing atmosphere (SR ≤ 0.8) conditions.

**Figure 3.** Comparison of different reaction mechanisms and models in staged air combustion: (**a**–**c**) are the predicted volume flow rates of NH3, HCN and NO, respectively, at the main combustion zone exit under the SR of 0.7.

The sums of NH3, HCN, and NO*<sup>x</sup>* at the main combustion zone and burnout zone exits in staged air combustion are shown in Figure 5. At the main combustion zone exit, the NO*<sup>x</sup>* emissions are low, while the sum of NH3, HCN, and NO*<sup>x</sup>* is quite high when the reaction temperature is set as 1200 ◦C. A large amount of NH3 and HCN remain in the primary combustion exhaust. With the secondary gas introduced, the remaining NH3 and HCN are almost entirely oxidized to NO*x*. Therefore, at the burnout zone exit, very little NH3 and HCN exist, and the sum of NH3, HCN, and NO*<sup>x</sup>* is approximately equal to the NO*<sup>x</sup>* emissions. The final NO*<sup>x</sup>* emissions depend on the sum of nitrogen compounds in the primary combustion exhaust. To limit the NO*<sup>x</sup>* emissions as much as possible by air staging, it is of great significance to obtain a minimum sum of NH3, HCN, and NO*<sup>x</sup>* in the main combustion zone, i.e., converting more fuel-N to N2 in terms of N balance. With the reaction temperature in the main combustion zone rising, the final NO*x* emissions decrease. When the temperature is higher than 1600 ◦C, the final NO*<sup>x</sup>* emissions attain a minimum level.

**Figure 4.** Conversion ratio of fuel-N to NO*x* in staged air combustion (reaction temperature in the burnout zone: 1100 ◦C): (**a**) main combustion zone exit; (**b**) burnout zone exit.

**Figure 5.** Sums of NH3, HCN and NO*<sup>x</sup>* at the exits in staged air combustion (reaction temperature in the burnout zone: 1100 ◦C).

Figure 6 illustrates the NO*<sup>x</sup>* emissions under the oxidizing atmosphere condition (SR ≥ 1) in air combustion. Considering the formation of the thermal NO*x*, a simulation of O2/Ar combustion was performed for comparison. Most of the fuel-N is easily oxidized to NO*<sup>x</sup>* by the excess O2. When the combustion temperature is higher than 1500 ◦C, the thermal NO*<sup>x</sup>* begins to be markedly produced. As the combustion temperature and SR rise, the emissions of the thermal NO*x* increase rapidly.

**Figure 6.** Comparison of NO*<sup>x</sup>* emissions between air combustion and O2/Ar combustion under oxidizing atmosphere conditions.

The effect of the reaction temperature in the burnout zone (*T*2) on the NO*<sup>x</sup>* emissions in air staging is shown in Figure 7. The reaction temperature in the main combustion zone (*T*1) is selected as 1500 ◦C. There is minimal difference in the final NO*<sup>x</sup>* CR when *T*<sup>2</sup> varies from 1100 to 1400 ◦C. In other words, *T*<sup>2</sup> nearly has no effect on the formation of NO*x*. Therefore, it is important to control the NO*<sup>x</sup>* formation and reduction in the main combustion zone instead of the burnout zone.

**Figure 7.** Effect of reaction temperature in the burnout zone on the NO*x* emissions in staged air combustion.

#### *4.3. Characteristics of NOx Emissions in Staged O2*/*CO2 Combustion*

Figure 8 presents the effects of the SR and combustion temperature on the NO*<sup>x</sup>* emissions in the staged O2/CO2 combustion. The variation trends of the NO*<sup>x</sup>* CRs are similar to those in the staged air combustion. However, a significant difference in the exact NO*<sup>x</sup>* emission value between O2/CO2 combustion and air combustion is found due to the existence of a great deal of CO2. In the O2/CO2 combustion, the CO2 concentration is so high that the chemical reaction 12 is observably facilitated [30,37]:

$$\text{CH} + \text{CO}\_2 \rightarrow \text{OH} + \text{CO} \tag{12}$$

Here, CO2 cannot be considered as an inert gas anymore. The chemical reaction 12 diminishes the H atom concentration and increases the concentration of OH radicals, which impacts the NO*<sup>x</sup>* formation and reduction strongly. Compared with the staged air combustion, the NO*x* CR at the main combustion zone exit increases markedly, and the range of the SR and combustion temperature (under which a significant amount of NH3 and HCN remain in the primary combustion exhaust) is narrow. At low temperatures (≤1400 ◦C), a significant amount of NH3 and HCN remain at the main combustion zone exit. Their sum rises rapidly with the decreasing SR and combustion temperature. Similarly, the lower final NO*<sup>x</sup>* emission levels appear at high temperatures (≥1500 ◦C), and the higher combustion temperature and the smaller SR lead to lower NO*x* emissions. Furthermore, the final NO*x* emissions in the staged O2/CO2 combustion are higher than those in the staged air combustion at the same high temperature and strong reducing atmosphere conditions. This conclusion is consistent with Mendiara et al.'s research results [32,36]. Because of the chemical reaction 12, the OH/H ratio increases, which is equivalent to providing an oxidizing agent in the combustion atmosphere. Therefore, the oxidation of NH3 and HCN to NO*<sup>x</sup>* is promoted.

**Figure 8.** Conversion ratio of fuel-N to NO*<sup>x</sup>* in staged O2/CO2 combustion (reaction temperature in the burnout zone: 1100 ◦C): (**a**) main combustion zone exit; (**b**) burnout zone exit.

Figure 9 compares the NO*<sup>x</sup>* emissions of staged air and O2/CO2 combustion under different atmosphere conditions. In O2/CO2 combustion, staged combustion is also able to decrease the NO*<sup>x</sup>* emissions enormously, but the emission reduction is less than that in air staging. Under the oxidizing atmosphere condition, the NO*<sup>x</sup>* emission levels are quite high in both air combustion and O2/CO2 combustion. Moreover, CO2 can reduce the O/H radical pool and tends to inhibit the NO*<sup>x</sup>* formation from fuel-N and, thus, the NO*<sup>x</sup>* emissions in O2/CO2 combustion are lower than those in air combustion. When under the reducing atmosphere condition, as the SR falls, the NO*<sup>x</sup>* CR in O2/CO2 combustion decreases while that in air combustion reduces first and then increases. The NO*<sup>x</sup>* CR of O2/CO2 combustion has a minimum of 7.1% at the SR of 0.5, while that of air combustion reaches a minimum of 4.7% at the SR of 0.7. These results denote that staged air combustion is more practical for limiting the NO*<sup>x</sup>* emissions than staged O2/CO2 combustion.

**Figure 9.** Comparison of NO*<sup>x</sup>* emissions between air combustion and O2/CO2 combustion under different atmosphere conditions.

Another interesting finding in Figure 9 is that there are two reverse trends for staged air combustion and O2/CO2 combustion at a smaller SR. The chemical reaction 12 can affect reducibility of combustion atmosphere strongly. At a smaller SR, a certain amount of NH3 and HCN remain in the primary combustion exhaust for staged air combustion when the temperature is not so high (here 1500 ◦C), then the NO*<sup>x</sup>* CR increases after these NH3 and HCN are oxidized to NO*<sup>x</sup>* by the OFA. While for staged O2/CO2 combustion, the atmosphere is much less reductive at the same SR; only a small amount of NH3 and HCN remain when the temperature is 1500 ◦C. Moreover, the smaller SR is, the more significant the effect of reducibility.

#### *4.4. Mechanism Analysis*

According to the ROP analysis, a reaction path diagram reflecting the main reaction pathways for the conversion of NH3 and HCN to NO or N2 in staged air and O2/CO2 combustion is proposed in Figure 10. The combustion temperature is 1600 ◦C and the SR is 0.7 during the calculation. The solid lines represent reaction pathways important in air combustion, while the dashed lines express those only significant in O2/CO2 combustion. It can be seen from the reaction path diagram that NO is directly reduced to N2 mainly through the following reactions:

$$\text{N} + \text{NO} \rightarrow \text{N}\_2 + \text{O} \tag{13}$$

$$\text{NH} + \text{NO} \rightarrow \text{N}\_2 + \text{OH} \tag{14}$$

$$\rm NCO + NO \rightarrow \rm N\_2 + CO\_2 \tag{15}$$

Besides, part of NO first forms nitrogen intermediates NNH and N2O, and they are then converted into N2 by:

$$\text{NH} + \text{NO} \rightarrow \text{NNH} + \text{O} \tag{16}$$

$$\text{NNH} + \text{H} \rightarrow \text{H}\_2 + \text{N}\_2 \tag{17}$$

$$\text{NNH} + \text{OH} \rightarrow \text{H}\_2\text{O} + \text{N}\_2\tag{18}$$

$$\text{NNH} \rightarrow \text{N}\_2 + \text{H} \tag{19}$$

$$\text{NH} + \text{NO} \rightarrow \text{N}\_2\text{O} + \text{H} \tag{20}$$

$$\rm NCO + NO \rightarrow \rm N\_2O + CO \tag{21}$$

$$\text{N}\_2\text{O} + \text{H} \rightarrow \text{N}\_2 + \text{OH} \tag{22}$$

$$\text{N}\_2\text{O}(+\text{M}) \rightarrow \text{N}\_2 + \text{O}(+\text{M})\tag{23}$$

**Figure 10.** Reaction path diagram for the fuel-N conversion under a high temperature and strong reducing atmosphere conditions in staged air and O2/CO2 combustion.

Major reactions for NH3 consumption are the interactions with H, O and OH radicals:

$$\rm NH\_3 + H \rightarrow NH\_2 + H\_2 \tag{24}$$

$$\text{NH}\_3 + \text{O} \rightarrow \text{NH}\_2 + \text{OH} \tag{25}$$

$$\text{NH}\_3 + \text{OH} \rightarrow \text{NH}\_2 + \text{H}\_2\text{O} \tag{26}$$

Main reactions for HCN removal are listed as follows:

$$\text{NHCN} + \text{OH} \rightarrow \text{NH}\_2 + \text{CO} \tag{27}$$

$$\text{CHCN} + \text{O} \rightarrow \text{NH} + \text{CO} \tag{28}$$

$$\text{HCN} + \text{O} \rightarrow \text{NCO} + \text{H} \tag{29}$$

$$\text{HCN} + \text{OH} \rightarrow \text{HOCN} + \text{H} \tag{30}$$

$$\text{HCN} + \text{OH} \rightarrow \text{HNO} + \text{H} \tag{31}$$

$$\text{CHCN} + \text{M} \rightarrow \text{H} + \text{CN} + \text{M} \tag{32}$$

$$\text{HCN} + \text{OH} \rightarrow \text{CN} + \text{H}\_2\text{O} \tag{33}$$

There are also some important nitrogen intermediates formed during the conversion of NH3 and HCN, such as NCO, NH, NH2, and HNO. Whether the fuel-N is finally converted into NO or N2 depends on the formation and evolution of these nitrogen intermediates, which are significantly affected by the presence of H, O and OH radicals in the reaction atmosphere. Determined by the combustion temperature and SR, different concentrations of H, O and OH radicals lead to different NO*x* emissions. Under the high temperature and strong reducing atmosphere conditions, the H atom concentration is increased, and the OH/H ratio and O/H ratio are decreased correspondingly. As a result, the reactions by which the nitrogen intermediates are oxidized to NO are inhibited, while those promoting the NO reduction to N2 are enhanced. For O2/CO2 combustion, reaction 12 is considered to be responsible for the impact of high CO2 concentration. It can compete with reaction 34 for H [30,37], which changes the concentrations of H, O and OH radicals in the reaction atmosphere.

$$\text{H} + \text{O}\_2 \rightarrow \text{O} + \text{OH} \tag{34}$$

Figure 11 gives the results of a first-order sensitivity analysis for N2 at the SR of 0.7 in air combustion. Here, the effect of temperature on N2 production is detected emphatically. The first-order sensitivity coefficients of some reactions are negative at a low temperature, while they become positive at a high temperature, which means that these reactions play important roles in the reduction of NO*x* to N2. With the increasing temperature, the function of these reactions switches from inhibiting N2 production to facilitating it. This is because the increasing rate constant of each reaction induced by the higher temperature changes the radical pool composition and size in the reaction atmosphere. Moreover, N2 is mostly sensitive to the reactions that generate or consume H and CH3 radicals under the high temperature and strong reducing atmosphere conditions. For instance, increasing the rate of reaction 34 will promote NO*<sup>x</sup>* reduction. Similarly, Figure 12 displays the first-order sensitivity analysis for N2 in O2/CO2 combustion. Due to reaction 12, some reactions for CH3 consumption become bottlenecks in N2 formation, besides reaction 34.

**Figure 11.** First-order sensitivity analysis for N2 at different temperatures in air combustion (SR = 0.7).

**Figure 12.** First-order sensitivity analysis for N2 at different temperatures in O2/CO2 combustion (SR = 0.7).

#### **5. Conclusions**

In this study, a methane flame doped with ammonia and hydrogen cyanide for fuel-N in a tandem-type tube furnace was simulated to investigate the characteristics of NO*<sup>x</sup>* emissions under a high temperature and strong reducing atmosphere conditions in staged air and O2/CO2 combustion by CHEMKIN. The effects of combustion temperature and stoichiometric ratio on the NO*<sup>x</sup>* emissions were examined, and the elementary steps for NO*x* formation and reduction at high temperatures were identified. The following conclusions can be drawn:

In both staged air and staged O2/CO2 combustion (SR < 1), the NO*<sup>x</sup>* CRs at the main combustion zone exit increase as the SR rises, and they are slightly affected by the combustion temperature. The NO*x* CRs at the burnout zone exit decrease with the increasing SR at low temperatures, and they are much higher than those at the main combustion zone exit. Here, a lot of nitrogen compounds remain in the exhaust of the main combustion zone and can be easily oxidized to NO*x* with the injection of secondary gas. Staged combustion can lower the NO*x* emission levels significantly, especially under a high temperature (≥1600 ◦C) and strong reducing atmosphere (SR ≤ 0.8) conditions. Increasing the combustion temperature under strong reducing atmosphere conditions can raise the H atom concentration and change the radical pool composition and size. Therefore, the reactions by which NO is reduced to N2 are facilitated. In addition, the increased OH/H ratio through reaction 12 offsets part of the reducibility in staged O2/CO2 combustion, resulting in the final NO*<sup>x</sup>* emissions in O2/CO2 combustion being higher than those in air combustion at the same high temperature and strong reducing atmosphere conditions.

**Author Contributions:** Conceptualization, S.W. and D.C.; methodology, S.W.; software, S.W. and Z.W.; validation, S.W. and X.S.; data curation, S.W. and D.C.; writing—original draft preparation, S.W. and D.C.; writing—review and editing, Z.W. and X.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work has been financially supported by the National Natural Science Foundation of China (Grant No. 51906202) and the Basic Research Program of Natural Science of Shaanxi Province (No. 2019JQ-809).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Nomenclature**


#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
