Co 3 O 4 / g-C 3 N 4 Hybrids for Gas-Phase Hg 0 Removal at Low Temperature

: The Co 3 O 4 / g-C 3 N 4 hybrids are constructed via the incipient wetness impregnation method by depositing Co 3 O 4 onto the exterior of g-C 3 N 4 , and then employed for Hg 0 capture within 60–240 ◦ C. The results show that the Co 3 O 4 / g-C 3 N 4 hybrid with a Co 3 O 4 content of 12 wt% performs optimally with the highest Hg 0 removal e ﬃ ciency of ~100% at or above 120 ◦ C. The high performances of the Co 3 O 4 / g-C 3 N 4 hybrids are probably attributed to the tight interfacial contact between Co 3 O 4 and g-C 3 N 4 , with its improved electron transfer, inferring that cobalt oxide and g-C 3 N 4 display a cooperative e ﬀ ect towards Hg 0 removal. NO and SO 2 shows a signiﬁcant suppressive inﬂuence on the mercury capture performance, plausibly owing to the competing adsorption and side reactions.


Introduction
Mercury, as a kind of heavy metal, is a persistent toxic substance of increasing concern worldwide [1,2]. The combustion of coal accounts for the great majority of the global anthropogenic mercury emissions [3]. Coal-fired mercury can be empirically classified into three main forms: elemental mercury (Hg 0 ), oxidized mercury (Hg 2+ ), and particulate mercury (Hg(p)). Hg 2+ and Hg(p) can be removed using wet scrubbers and fabric filters, respectively [4]. However, elemental mercury (Hg 0 ), the predominant speciation of coal-derived mercury, is hardly captured by the available pollutant control facilities because it is highly volatile, chemically stable, and water insoluble [5]. Catalytic oxidation and adsorption is considered to be the most effective and viable method for the reduction of Hg 0 emission [6]. Activated carbon is frequently used for mercury emission control in power plants [7]. Nevertheless, the high operation cost, low mercury capacity, and negative effect on fly ash quality has restrained its widespread application in coal-fired power plants [8]. Currently, transition-metal oxides, such as CuO [9], MnO 2 [10], Fe 2 O 3 [11], and Co 3 O 4 [12], have been recognized as potential sorbents or catalysts for Hg 0 removal because of their excellent redox abilities and lower capital cost. CeO 2 , as a rare-earth oxide, has also been employed as an additive, active ingredient, and carrier in the synthesis of composites for mercury capture, attributed to the CeO 2 /Ce 2 O 3 redox couple and structure defect [13].
Metal oxides loading onto the exterior of nano-materials is a type of catalyst with good adsorption as well as redox ability [14]. Thus, these catalysts could be potential candidates for capturing Hg 0 from coal combustion flue gas. Graphite-like carbon nitride (g-C 3 N 4 ), the most stable allotrope of versatile CN structures, has been widely studied in many scientific fields because of its distinct electronic structure, thermal stability, ample source as well as simple synthesis route [15]. Two-dimensional g-C 3 N 4 nanosheet possesses the advantage of providing more anchoring sites for bare metal oxide

Characterization Analysis
The FESEM image and photo of the CNNS are shown in Figure 2. CNNS displays a light-yellow color (the inset in Figure 2a) and it consists of many lamellar structures with lateral sizes of < 1 μm ( Figure 2a) and thickness of dozens of nanometers ( Figure 2b). The CNNS displays transparent characteristics, indicating the super-thin sheet-like morphology of the CNNS with ~1-10 nm in thickness and ~1-3 μm in size (Figure 3a). The TEM image and the selected area electron diffraction (SAED) analysis of the 12Co3O4/CNNS are displayed in Figure 3b. The lattice fringes of Co3O4 are closely surrounded by g-C3N4. The intimate interfacial contact implies the good interaction between Co3O4 and g-C3N4, which is beneficial for electron transfer and, thus, promoting Hg 0 oxidation [26]. The diffraction spots with a distance of 0.205 and 0.429 nm are ascribed to the reflection of the Co3O4 (400) and Co3O4 (111) lattice plane, respectively. The XRD spectra of the pure and Co3O4-modified CNNS are displayed in Figure 4. The intense peaks at ~27.7° detected in all specimens belong to the reflection of the (002) crystal face of g-C3N4 (JCPDS no. 87-1526) [27]. The signals at 31.1°, 36.7°, 44.8°, 59.4°, and 65.1° are assigned to the diffraction of Co3O4 (JCPDS no. 42-1467) [28]. With the increment of the Co3O4 content, its feature peaks gradually increase in intensity, suggesting that Co3O4 has been successfully deposited on g-C3N4 surface.

Characterization Analysis
The FESEM image and photo of the CNNS are shown in Figure 2. CNNS displays a light-yellow color (the inset in Figure 2a) and it consists of many lamellar structures with lateral sizes of <1 µm ( Figure 2a) and thickness of dozens of nanometers ( Figure 2b). The CNNS displays transparent characteristics, indicating the super-thin sheet-like morphology of the CNNS with~1-10 nm in thickness and~1-3 µm in size (Figure 3a). The TEM image and the selected area electron diffraction (SAED) analysis of the 12Co 3 O 4 /CNNS are displayed in Figure 3b. The lattice fringes of Co 3 O 4 are closely surrounded by g-C 3 N 4 . The intimate interfacial contact implies the good interaction between Co 3 O 4 and g-C 3 N 4 , which is beneficial for electron transfer and, thus, promoting Hg 0 oxidation [26]. The diffraction spots with a distance of 0.205 and 0.429 nm are ascribed to the reflection of the Co 3 O 4 (400) and Co 3 O 4 (111) lattice plane, respectively. The XRD spectra of the pure and Co 3 O 4 -modified CNNS are displayed in Figure 4. The intense peaks at~27.7 • detected in all specimens belong to the reflection of the (002) crystal face of g-C 3 N 4 (JCPDS no. 87-1526) [27]. The signals at 31.1 • , 36.7 • , 44.8 • , 59.4 • , and 65.1 • are assigned to the diffraction of Co 3 O 4 (JCPDS no. 42-1467) [28]. With the increment of the Co 3 O 4 content, its feature peaks gradually increase in intensity, suggesting that Co 3 O 4 has been successfully deposited on g-C 3 N 4 surface.

Characterization Analysis
The FESEM image and photo of the CNNS are shown in Figure 2. CNNS displays a light-yellow color (the inset in Figure 2a) and it consists of many lamellar structures with lateral sizes of < 1 μm ( Figure 2a) and thickness of dozens of nanometers ( Figure 2b). The CNNS displays transparent characteristics, indicating the super-thin sheet-like morphology of the CNNS with ~1-10 nm in thickness and ~1-3 μm in size (Figure 3a). The TEM image and the selected area electron diffraction (SAED) analysis of the 12Co3O4/CNNS are displayed in Figure 3b. The lattice fringes of Co3O4 are closely surrounded by g-C3N4. The intimate interfacial contact implies the good interaction between Co3O4 and g-C3N4, which is beneficial for electron transfer and, thus, promoting Hg 0 oxidation [26]. The diffraction spots with a distance of 0.205 and 0.429 nm are ascribed to the reflection of the Co3O4 (400) and Co3O4 (111) lattice plane, respectively. The XRD spectra of the pure and Co3O4-modified CNNS are displayed in Figure 4. The intense peaks at ~27.7° detected in all specimens belong to the reflection of the (002) crystal face of g-C3N4 (JCPDS no. 87-1526) [27]. The signals at 31.1°, 36.7°, 44.8°, 59.4°, and 65.1° are assigned to the diffraction of Co3O4 (JCPDS no. 42-1467) [28]. With the increment of the Co3O4 content, its feature peaks gradually increase in intensity, suggesting that Co3O4 has been successfully deposited on g-C3N4 surface.     The nitrogen isotherms of the pure and Co3O4-modified CNNS are presented in Figure 5. Hysteresis loops were observed in all specimens at higher relative pressure. The nitrogen uptakes at lower relative pressures were fairly less, implying that enormous mesoporous structures presented on the catalyst surface [29]. The pure CNNS had a large BET surface area of 109 m 2 /g and large average pore size of 19 nm (Table 1). After incorporating Co3O4, they all reduced significantly. The BET surface area and average pore size of xCo3O4/CNNS dropped to 27-42 m 2 /g and 10-12 nm, respectively. The probable reason for this is that parts of the mesoporous structures of the CNNS were filled or blocked by Co3O4 grains [30].    The nitrogen isotherms of the pure and Co3O4-modified CNNS are presented in Figure 5. Hysteresis loops were observed in all specimens at higher relative pressure. The nitrogen uptakes at lower relative pressures were fairly less, implying that enormous mesoporous structures presented on the catalyst surface [29]. The pure CNNS had a large BET surface area of 109 m 2 /g and large average pore size of 19 nm (Table 1). After incorporating Co3O4, they all reduced significantly. The BET surface area and average pore size of xCo3O4/CNNS dropped to 27-42 m 2 /g and 10-12 nm, respectively. The probable reason for this is that parts of the mesoporous structures of the CNNS were filled or blocked by Co3O4 grains [30].  The nitrogen isotherms of the pure and Co 3 O 4 -modified CNNS are presented in Figure 5. Hysteresis loops were observed in all specimens at higher relative pressure. The nitrogen uptakes at lower relative pressures were fairly less, implying that enormous mesoporous structures presented on the catalyst surface [29]. The pure CNNS had a large BET surface area of 109 m 2 /g and large average pore size of 19 nm (Table 1). After incorporating Co 3 O 4 , they all reduced significantly. The BET surface area and average pore size of xCo 3 O 4 /CNNS dropped to 27-42 m 2 /g and 10-12 nm, respectively. The probable reason for this is that parts of the mesoporous structures of the CNNS were filled or blocked by Co 3 O 4 grains [30]. The FTIR profiles of fresh CNNS and xCo3O4/CNNS are displayed in Figure 6a. The two sharp signals at ~892 and ~807 cm −1 are in accord with the feature breathing pattern of the triazine loop units stemming from polymerized C-N heterocycles [31]. The set of peaks between ~1109 and ~1745 cm −1 belong to the aromatic C-N stretch oscillation [32]; the weak bands at ~2143 cm −1 correspond to the cyano C≡N terminal groups [33]. The peaks at ~2352 cm −1 belong to the O=C=O asymmetrical stretch  The FTIR profiles of fresh CNNS and xCo 3 O 4 /CNNS are displayed in Figure 6a. The two sharp signals at~892 and~807 cm −1 are in accord with the feature breathing pattern of the triazine loop units stemming from polymerized C-N heterocycles [31]. The set of peaks between~1109 and~1745 cm −1 belong to the aromatic C-N stretch oscillation [32]; the weak bands at~2143 cm −1 correspond to the cyano C≡N terminal groups [33]. The peaks at~2352 cm −1 belong to the O=C=O asymmetrical stretch oscillation of the carbon dioxide adhered to the catalyst exterior [34]. The broad bands between 2927 and 3688 cm −1 relate to the N-H stretch oscillation of unpolymerized -NH 2 functions and the O-H stretch oscillation of the H 2 O molecules adhered to the catalyst exterior [35]. As depicted in Figure 6b, the main feature peaks of the g-C 3 N 4 can be distinctly detected in the spent 12Co 3 O 4 /CNNS, indicating that the atomic structures of the 12Co 3 O 4 /CNNS retain the same during Hg 0 removal reactions. The FTIR profiles of fresh CNNS and xCo3O4/CNNS are displayed in Figure 6a. The two sharp signals at ~892 and ~807 cm −1 are in accord with the feature breathing pattern of the triazine loop units stemming from polymerized C-N heterocycles [31]. The set of peaks between ~1109 and ~1745 cm −1 belong to the aromatic C-N stretch oscillation [32]; the weak bands at ~2143 cm −1 correspond to the cyano C≡N terminal groups [33]. The peaks at ~2352 cm −1 belong to the O=C=O asymmetrical stretch oscillation of the carbon dioxide adhered to the catalyst exterior [34]. The broad bands between 2927 and 3688 cm −1 relate to the N-H stretch oscillation of unpolymerized -NH2 functions and the O-H stretch oscillation of the H2O molecules adhered to the catalyst exterior [35]. As depicted in Figure  6b, the main feature peaks of the g-C3N4 can be distinctly detected in the spent 12Co3O4/CNNS, indicating that the atomic structures of the 12Co3O4/CNNS retain the same during Hg 0 removal reactions. The surface elemental valences of 12Co3O4/CNNS before and after reaction were examined by XPS, as presented in Figure 7. The bands at ~284.6 and ~285.2 eV relate to the sp 2 -bonded carbon atoms, while the peaks at ~287.6 and ~288.2 eV are associated with the N-C=N structures [36]. The peaks at ~398.0 and ~398.4 eV belong to the hybridized secondary nitrogen (C-N=C). The bands at ~399.0 and ~399.2 eV are related to the hybridized tertiary nitrogen (N-(C)3). The peaks at ~400.0 and ~400.4 eV correspond to the -NH2 groups [37]. What is more, the C 1s and N 1s binding energies of the used 12Co3O4/CNNS are smaller than those of the fresh one. This suggests that Hg 0 probably donates electrons to g-C3N4 during mercury removal reactions. With respect to the Co 2p spectrum, the bands at ~782.0 and ~782.8 eV are supposed to be the reflection of the Co 3+ in the Co3O4, while the peaks at ~786.4 and ~787.4 eV relate to the signals of the Co 2+ in the Co3O4 [38]. The ratio of Co 3+ /Co 2+ for the fresh specimen is ~0.83. It slightly decreases to ~0.75 for the spent specimen. This indicates that a fraction of the Co 3+ cations transferred into the Co 2+ cations after Hg 0 oxidation. The Co 3+ cations are involved in the mercury oxidation reactions. The bands at ~531.8 eV belong to the chemisorbed The surface elemental valences of 12Co 3 O 4 /CNNS before and after reaction were examined by XPS, as presented in Figure 7. The bands at~284.6 and~285.2 eV relate to the sp 2 -bonded carbon atoms, while the peaks at~287.6 and~288.2 eV are associated with the N-C=N structures [36]. The peaks at~398.0 and~398.4 eV belong to the hybridized secondary nitrogen (C-N=C). The bands at 399.0 and~399.2 eV are related to the hybridized tertiary nitrogen (N-(C) 3 ). The peaks at~400.0 and 400.4 eV correspond to the -NH 2 groups [37]. What is more, the C 1s and N 1s binding energies of the used 12Co 3 O 4 /CNNS are smaller than those of the fresh one. This suggests that Hg 0 probably donates electrons to g-C 3 N 4 during mercury removal reactions. With respect to the Co 2p spectrum, the bands at~782.0 and~782.8 eV are supposed to be the reflection of the Co 3+ in the Co 3 O 4 , while the peaks at 786.4 and~787.4 eV relate to the signals of the Co 2+ in the Co 3 O 4 [38]. The ratio of Co 3+ /Co 2+ for the fresh specimen is~0.83. It slightly decreases to~0.75 for the spent specimen. This indicates that a fraction of the Co 3+ cations transferred into the Co 2+ cations after Hg 0 oxidation. The Co 3+ cations are involved in the mercury oxidation reactions. The bands at~531.8 eV belong to the chemisorbed oxygen (O α ) on the catalyst exterior, while the peaks at~533.2 eV are linked to the oxygen (O β ) in the -OH group or C-O bond [39]. The percentage of the chemisorbed oxygen increased from~54.7 to~61.3% for the 12Co 3 O 4 /CNNS before and after the reaction, correspondingly, which infers that the gaseous O 2 from the feed gas may replenish the depleted chemisorbed oxygen during the Hg 0 oxidation processes. The Hg 4f spectra displays two feature peaks at~101.4 and~103.6 eV, which implies the generation of HgO [40]. Moreover, the absence of the feature peak of elemental mercury at 99.9 eV indicates that Hg 0 adsorption over 12Co 3 O 4 /CNNS is governed by chemisorptions [41].
Processes 2019, 7, x FOR PEER REVIEW 6 of 11 oxygen (Oα) on the catalyst exterior, while the peaks at ~533.2 eV are linked to the oxygen (Oβ) in the -OH group or C-O bond [39]. The percentage of the chemisorbed oxygen increased from ~54.7 to ~61.3% for the 12Co3O4/CNNS before and after the reaction, correspondingly, which infers that the gaseous O2 from the feed gas may replenish the depleted chemisorbed oxygen during the Hg 0 oxidation processes. The Hg 4f spectra displays two feature peaks at ~101.4 and ~103.6 eV, which implies the generation of HgO [40]. Moreover, the absence of the feature peak of elemental mercury at 99.9 eV indicates that Hg 0 adsorption over 12Co3O4/CNNS is governed by chemisorptions [41].

Impact of Loading Value
The performances of the pure and Co3O4-modified CNNS with respect to elemental mercury removal at 120 °C in 5% O2/N2 over 90 min are presented in Figure 8. The pristine CNNS shows good Hg 0 capture ability, plausibly because of the distinct C-N structure and considerable surface area. The equilibrium Hg 0 removal efficiency is ~59.0%. Incorporating Co3O4 with g-C3N4 could remarkably enhance the Hg 0 removal ability. The Hg 0 sorption rate at the initial stage and the Hg 0 removal efficiency both increase after the addition of Co3O4. The mercury conversion rises from ~59.0 to ~86.0% as the Co3O4 loading value elevates from 0 to 8 wt%. The 12Co3O4/CNNS performs the best with mercury conversion reaching ~100%, indicating that the Co3O4 modification contributes to Hg 0 removal by the introduction of added reactive sites. Moreover, the tight interfacial interaction of

Impact of Loading Value
The performances of the pure and Co 3 O 4 -modified CNNS with respect to elemental mercury removal at 120 • C in 5% O 2 /N 2 over 90 min are presented in Figure 8. The pristine CNNS shows good Hg 0 capture ability, plausibly because of the distinct C-N structure and considerable surface area. The equilibrium Hg 0 removal efficiency is~59.0%. Incorporating Co 3 O 4 with g-C 3 N 4 could remarkably enhance the Hg 0 removal ability. The Hg 0 sorption rate at the initial stage and the Hg 0 removal efficiency both increase after the addition of Co 3 O 4 . The mercury conversion rises from~59.0 to~86.0% as the Co 3 O 4 loading value elevates from 0 to 8 wt%. The 12Co 3 O 4 /CNNS performs the best with mercury conversion reaching~100%, indicating that the Co 3 O 4 modification contributes to Hg 0 removal by the introduction of added reactive sites. Moreover, the tight interfacial interaction of Co 3 O 4 and g-C 3 N 4 , with promoted charge transfer mobility, can facilitate redox reactions, which could enhance mercury conversion. Nevertheless, the redox ability would decline with further incremental Co 3 O 4 content. The mercury conversion subtly drops to~97.5% when loading value exceeds 12 wt%, presumably owing to the significant reduction of the surface area. Thus, the best Co 3 O 4 content for xCo 3 O 4 /CNNS is 12 wt%. What is more, bare Co 3 O 4 exhibits a poor performance towards mercury adsorption with a mercury conversion of only~36.3%, probably owing to its larger grain size and lower surface area. Thus, it can be concluded that Co 3 O 4 and g-C 3 N 4 display a cooperative effect towards Hg 0 removal. Co3O4 and g-C3N4, with promoted charge transfer mobility, can facilitate redox reactions, which could enhance mercury conversion. Nevertheless, the redox ability would decline with further incremental Co3O4 content. The mercury conversion subtly drops to ~97.5% when loading value exceeds 12 wt%, presumably owing to the significant reduction of the surface area. Thus, the best Co3O4 content for xCo3O4/CNNS is 12 wt%. What is more, bare Co3O4 exhibits a poor performance towards mercury adsorption with a mercury conversion of only ~36.3%, probably owing to its larger grain size and lower surface area. Thus, it can be concluded that Co3O4 and g-C3N4 display a cooperative effect towards Hg 0 removal.

Impact of Reaction Temperature
The mercury conversion of the 12Co3O4/CNNS at 60-240 °C in 5% O2/N2 is presented in Figure  9. It is found that temperature could significantly affect mercury removal performance. Fairly low mercury oxidation activity was exhibited at 60 °C by 12Co3O4/CNNS, with a mercury conversion of only ~33.3%. The performance of mercury oxidation can be remarkably enhanced when temperature goes up to 90 °C, with mercury conversion swiftly rising to ~98.4%. A mercury conversion of 100% can be reached at temperature above or at 120 °C. The lower reaction activity is attributed to the lower reaction rate at lower temperatures, while higher temperature facilitates chemisorption processes because of the decreased activation energy barrier. As displayed in Figure 10, CNNS shows an unstable performance, with mercury conversion gradually dropping from ~77.2 to ~47.3% as time passed likely attributed to the consumption of the active sorption sites. In contrast, 12Co3O4/CNNS performs stably, with mercury conversion rapidly reaching ~100% within ~42 min and then leveling off as time passes, plausibly owing to its prominent redox ability.

Impact of Reaction Temperature
The mercury conversion of the 12Co 3 O 4 /CNNS at 60-240 • C in 5% O 2 /N 2 is presented in Figure 9. It is found that temperature could significantly affect mercury removal performance. Fairly low mercury oxidation activity was exhibited at 60 • C by 12Co 3 O 4 /CNNS, with a mercury conversion of only~33.3%. The performance of mercury oxidation can be remarkably enhanced when temperature goes up to 90 • C, with mercury conversion swiftly rising to~98.4%. A mercury conversion of 100% can be reached at temperature above or at 120 • C. The lower reaction activity is attributed to the lower reaction rate at lower temperatures, while higher temperature facilitates chemisorption processes because of the decreased activation energy barrier. As displayed in Figure 10, CNNS shows an unstable performance, with mercury conversion gradually dropping from~77.2 to~47.3% as time passed likely attributed to the consumption of the active sorption sites. In contrast, 12Co 3 O 4 /CNNS performs stably, with mercury conversion rapidly reaching~100% within~42 min and then leveling off as time passes, plausibly owing to its prominent redox ability.
Co3O4 and g-C3N4, with promoted charge transfer mobility, can facilitate redox reactions, which could enhance mercury conversion. Nevertheless, the redox ability would decline with further incremental Co3O4 content. The mercury conversion subtly drops to ~97.5% when loading value exceeds 12 wt%, presumably owing to the significant reduction of the surface area. Thus, the best Co3O4 content for xCo3O4/CNNS is 12 wt%. What is more, bare Co3O4 exhibits a poor performance towards mercury adsorption with a mercury conversion of only ~36.3%, probably owing to its larger grain size and lower surface area. Thus, it can be concluded that Co3O4 and g-C3N4 display a cooperative effect towards Hg 0 removal.

Impact of Reaction Temperature
The mercury conversion of the 12Co3O4/CNNS at 60-240 °C in 5% O2/N2 is presented in Figure  9. It is found that temperature could significantly affect mercury removal performance. Fairly low mercury oxidation activity was exhibited at 60 °C by 12Co3O4/CNNS, with a mercury conversion of only ~33.3%. The performance of mercury oxidation can be remarkably enhanced when temperature goes up to 90 °C, with mercury conversion swiftly rising to ~98.4%. A mercury conversion of 100% can be reached at temperature above or at 120 °C. The lower reaction activity is attributed to the lower reaction rate at lower temperatures, while higher temperature facilitates chemisorption processes because of the decreased activation energy barrier. As displayed in Figure 10, CNNS shows an unstable performance, with mercury conversion gradually dropping from ~77.2 to ~47.3% as time passed likely attributed to the consumption of the active sorption sites. In contrast, 12Co3O4/CNNS performs stably, with mercury conversion rapidly reaching ~100% within ~42 min and then leveling off as time passes, plausibly owing to its prominent redox ability.

Impact of Flue Gas
Nitrogen monoxide and sulfur dioxide are the predominant acidic gas components in coalderived flue gas. The influences of NO and SO2 on the mercury conversion of 12Co3O4/CNNS at 120 °C is shown in Figure 11. It is found that NO and SO2 both show detrimental impacts on Hg 0 removal processes. The mercury conversion of 12Co3O4/CNNS was significantly reduced to ~44.6 and ~42.7% after adding 800 ppm NO and 1200 ppm SO2 into the feed gas, respectively. This suggests that NO or SO2 molecules and Hg 0 vapor could be competitively adsorbed on the surface of 12Co3O4/CNNS [9]. In addition, the adsorbed SO2 may react with Co3O4 to produce thermally-stable metal sulfates, i.e., cobalt sulfate, leading to the decline of the active phases [42]. Thus, the competitive adsorption and side reactions contribute to the remarkable suppressive influences of NO and SO2 towards Hg 0 adsorption. Figure 11. Impact of NO and SO2 on the mercury conversion of 12Co3O4/CNNS.

Mercury Capture Mechanism
In summary, according to the XPS analysis, Co3O4 serves as the main reactive site for elemental mercury capture. The combination of Co3O4 with g-C3N4 nanosheet increased the catalyst surface area as well as enhancing the redox capability of the catalyst, which is beneficial for the oxidation removal of elemental mercury. Additionally, the production of the Co3O4/g-C3N4 hybrids reduced the potential energy barrier and boosted the charge transfer mobility, which facilitates elemental mercury removal. Hence, the Hg 0 removal processes using Co3O4/g-C3N4 hybrids could be summarized into three stages. (i) The gas-phase Hg 0 and O2 adhere on the catalyst exterior to produce adhered mercury (Hg 0 ad) and chemisorbed oxygen (Oad). (ii) The Hg 0 ad is oxidized into Hg 2+ by the Co 3+ cations in Co3O4

Impact of Flue Gas
Nitrogen monoxide and sulfur dioxide are the predominant acidic gas components in coal-derived flue gas. The influences of NO and SO 2 on the mercury conversion of 12Co 3 O 4 /CNNS at 120 • C is shown in Figure 11. It is found that NO and SO 2 both show detrimental impacts on Hg 0 removal processes. The mercury conversion of 12Co 3 O 4 /CNNS was significantly reduced to~44.6 and~42.7% after adding 800 ppm NO and 1200 ppm SO 2 into the feed gas, respectively. This suggests that NO or SO 2 molecules and Hg 0 vapor could be competitively adsorbed on the surface of 12Co 3 O 4 /CNNS [9]. In addition, the adsorbed SO 2 may react with Co 3 O 4 to produce thermally-stable metal sulfates, i.e., cobalt sulfate, leading to the decline of the active phases [42]. Thus, the competitive adsorption and side reactions contribute to the remarkable suppressive influences of NO and SO 2 towards Hg 0 adsorption.

Impact of Flue Gas
Nitrogen monoxide and sulfur dioxide are the predominant acidic gas components in coalderived flue gas. The influences of NO and SO2 on the mercury conversion of 12Co3O4/CNNS at 120 °C is shown in Figure 11. It is found that NO and SO2 both show detrimental impacts on Hg 0 removal processes. The mercury conversion of 12Co3O4/CNNS was significantly reduced to ~44.6 and ~42.7% after adding 800 ppm NO and 1200 ppm SO2 into the feed gas, respectively. This suggests that NO or SO2 molecules and Hg 0 vapor could be competitively adsorbed on the surface of 12Co3O4/CNNS [9]. In addition, the adsorbed SO2 may react with Co3O4 to produce thermally-stable metal sulfates, i.e., cobalt sulfate, leading to the decline of the active phases [42]. Thus, the competitive adsorption and side reactions contribute to the remarkable suppressive influences of NO and SO2 towards Hg 0 adsorption. Figure 11. Impact of NO and SO2 on the mercury conversion of 12Co3O4/CNNS.

Mercury Capture Mechanism
In summary, according to the XPS analysis, Co3O4 serves as the main reactive site for elemental mercury capture. The combination of Co3O4 with g-C3N4 nanosheet increased the catalyst surface area as well as enhancing the redox capability of the catalyst, which is beneficial for the oxidation removal of elemental mercury. Additionally, the production of the Co3O4/g-C3N4 hybrids reduced the potential energy barrier and boosted the charge transfer mobility, which facilitates elemental mercury removal. Hence, the Hg 0 removal processes using Co3O4/g-C3N4 hybrids could be summarized into three stages. (i) The gas-phase Hg 0 and O2 adhere on the catalyst exterior to produce adhered mercury (Hg 0 ad) and chemisorbed oxygen (Oad). (ii) The Hg 0 ad is oxidized into Hg 2+ by the Co 3+ cations in Co3O4

Mercury Capture Mechanism
In summary, according to the XPS analysis, Co 3 O 4 serves as the main reactive site for elemental mercury capture. The combination of Co 3 O 4 with g-C 3 N 4 nanosheet increased the catalyst surface area as well as enhancing the redox capability of the catalyst, which is beneficial for the oxidation removal of elemental mercury. Additionally, the production of the Co 3 O 4 /g-C 3 N 4 hybrids reduced the potential energy barrier and boosted the charge transfer mobility, which facilitates elemental mercury removal. Hence, the Hg 0 removal processes using Co 3 O 4 /g-C 3 N 4 hybrids could be summarized into three stages. (i) The gas-phase Hg 0 and O 2 adhere on the catalyst exterior to produce adhered mercury (Hg 0 ad ) and chemisorbed oxygen (O ad ). (ii) The Hg 0 ad is oxidized into Hg 2+ by the Co 3+ cations in Co 3 O 4 (Hg 0 ad + 2Co 3+ → Hg 2+ + 2Co 2+ ). The produced Hg 2+ cations react with O ad or the lattice oxygen (O lat ) of Co 3 O 4 to generate HgO, which is then captured on the g-C 3 N 4 exterior. (iii) The oxygen in the feed gas can refresh the depleted O ad or O lat , and oxidize Co 2+ into Co 3+ to rebirth the active phase