Selective Catalytic Reduction of NO by NH₃ and SO_x Poisoning Mechanisms on Mn₃O₄ Catalysts: A Density Functional Investigation

Abstract

Mn-based oxides are promising catalysts for the selective catalytic reduction (SCR) of NO_x by NH₃ at low temperatures. However, fundamental NH₃-SCR mechanisms and resistance mechanisms against SO_x remain controversial. This study employed density functional theory (DFT) calculations to explore the intrinsic mechanisms of NH₃-SCR and SO_x poisoning on Mn₃O₄(001). Both NH₃ and NO adsorb atop the surface Mn site (the Lewis acid site). In contrast to the traditional Langmuir–Hinshelwood (L-H) mechanism in which gaseous NO is first oxidized to form adsorbed nitrites or nitrates and then react with adsorbed NH_x species to produce H₂O and N₂, a new potential L-H pathway is proposed that involves gaseous NO first adsorbing and then reacting with NH* to generate the key intermediate NHNO*, followed by the formation of H₂O and N₂. This L-H pathway is more efficient as it bypasses the NO oxidation step and is more selective for N₂ formation by avoiding N₂O production. In addition, the L-H mechanism is more favorable than the Eley–Rideal (E-R) mechanism because of the lower free energy profile. SO₂ exhibits limited poisoning effects, whereas SO₃ strongly poisons the Mn₃O₄(001) surface by occupying adsorption sites, hindering intermediate formation and producing ammonium bisulfate.

1. Introduction

The selective catalytic reduction (SCR) of nitrogen oxides (NO_x) by ammonia (NH₃) is a pivotal technology for mitigating NO_x emissions, which are notorious for their contribution to acid rain, photochemical smog, and respiratory health issues [1,2,3,4,5,6,7,8,9]. SCR catalysts are employed to facilitate the conversion of nitrogen oxides (NO_x) into nitrogen (N₂) and water (H₂O), typically through a redox reaction with ammonia (NH₃) as the reductant: 4NH₃ + 4NO + O₂ → 4N₂ + 6H₂O. The efficiency and durability of this process are heavily dependent on the performance of the catalyst used. Traditionally, V₂O₅-WO₃ (MoO₃)/TiO₂ has been the predominant commercial catalyst for NH₃-SCR due to its high activity and stability at temperatures of 300–400 °C [7,8,10,11,12]. However, the placement of these catalysts upstream in flue gas treatment processes can expose them to dust, sulfur dioxide (SO₂), and other impurities, which can significantly diminish their denitration efficiency and lead to catalyst poisoning or deactivation over time. In response to these challenges, there is a growing trend toward integrating SCR catalysts in low-dust arrangements, where the inlet flue gas temperature is approximately 200 °C [7,8,13,14,15,16]. This shift highlights the importance of developing catalysts that exhibit robust catalytic activity at lower temperatures, and Mn-based oxide catalysts have shown promise in this regard due to their enhanced performance in the low-temperature range [2,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38].

Mn₃O₄ has garnered considerable attention due to its tetragonally distorted spinel structure and complex canted spin configuration, making it an intriguing subject for both experimental and theoretical studies [27,28,39,40,41,42,43]. Kapteijn et al. [43] reported that the SCR activity of Mn oxides decreased in the order of MnO₂ > Mn₅O₈> Mn₂O₃ > Mn₃O₄. Despite MnO₂ exhibiting the highest activity per unit surface area, it is more susceptible to sulfur poisoning due to its high surface active oxygen content, which reacts readily with SO₂. On the other hand, Mn₃O₄, though displaying lower inherent NO oxidizing ability due to the reduced Mn valence state, offers promise as a low-temperature SCR deNO_x catalyst with enhanced SO₂ resistance. Previous studies highlighted by Kang et al. [44] and Xie et al. [45] have emphasized the presence of Mn₃O₄ as a significant component in active catalysts, suggesting its potential when modified appropriately. Mn₃O₄ offers a promising balance for enhancing both catalytic activity and durability. While not as extensively studied as other manganese oxides, an in-depth examination of the surface chemistry in SCR de-NO_x processes using Mn₃O₄ could greatly advance our molecular-level understanding of Mn-based catalysts, particularly at low temperatures.

The NH₃-SCR mechanism at low temperatures over MnO_x-based catalysts has been extensively studied both experimentally and theoretically [7,8,46,47]. The widely accepted reaction pathway involves adsorbed NH_x species reacting either with adsorbed nitrites/nitrates via the Langmuir–Hinshelwood (L-H) mechanism [30,44,48] or directly with gaseous NO and NO₂ through the Eley–Rideal (E-R) mechanism [43,49,50], forming NH_x−NO_x intermediates that decompose into N₂ and H₂O. In the L-H mechanism, reactions occur between two adsorbed species. For instance, Wu et al. [51] demonstrated that the SCR de-NO_x reaction on FeO_x-MnO_x/TiO₂ catalysts follows the L-H mechanism, where stable bidentate nitrates react with adsorbed NH₃ to form less stable monodentate nitrates and NH₄⁺, which subsequently react with NO to produce N₂ and H₂O. Similar findings have been reported in other studies [52,53]. Kapteijn et al. [43] proposed a model for pure MnO_x, where adsorbed NH₃ is progressively dehydrogenated by surface oxygen. Qi et al. [30] observed that gaseous NH₃ adsorbed on MnO_x-CeO₂ forms NH₃, NH₂, and OH species, with NO oxidized to NO₂, which then reacts with adsorbed NH₃ to form nitrates and nitrites that decompose into N₂ and H₂O. Conversely, Marbán et al. [49,50] suggested that SCR reactions over carbon-supported Mn₃O₄ catalysts proceed via the E-R mechanism, where surface-activated NH₃ reacts primarily with gaseous NO₂ and, to a lesser extent, NO. Xiong et al. [42] also found that the E-R mechanism dominates in low-temperature SCR reactions over Mn₃O₄ and (Cu_1.0Mn_2.0)_1–δO₄ spinels. Using DFT calculations, Yang et al. [41] investigated the SCR process on Mn₃O₄ (110) surfaces, revealing that NH₃ adsorbs on Lewis acid sites and dehydrogenates into NH₂* with the assistance of lattice oxygen. NH₂* then reacts with gaseous NO to form NH₂NO, which tends to convert into N₂O rather than N₂, with residual H atoms interacting with O₂ to form H₂O. Despite extensive research and the potential of Mn-based oxide catalysts, the fundamental NH₃-SCR mechanism at low temperatures remains controversial. Additionally, N₂ selectivity is a critical factor in de-NO_x performance, as side reactions can produce N₂O, a potent greenhouse gas. N₂O formation can occur through several pathways: (1) 3NO → N₂O + NO₂ [54], (2) 2NH₃ + 2O₂ → N₂O + 3H₂O [55], (3) 4NO + 4NH₃ + 3O₂ → 4N₂O + 6H₂O [43,56], and (4) 4NO₂ + 4NH₃ + O₂ → 4N₂O + 6H₂O [57,58]. Reaction intermediates significantly influence product distribution, making it essential to understand the pathways leading to nonselective NH₃ oxidation. These findings underscore the need for a reassessment of the NH₃-SCR mechanism on Mn₃O₄ surfaces to optimize N₂ selectivity and minimize N₂O formation.

In practical applications, residual SO_x (<100 ppm) persists in flue gas post-desulfurization, adversely affecting catalyst performance [7,8,59]. SO_x induces the formation of ammonium sulfate, which deposits on catalyst surfaces, deactivating them by blocking active sites. Metal oxide catalysts, particularly Mn-based ones, are vulnerable to deactivation due to ammonium sulfate and metal sulfates, which either block or destroy active sites [27,52,60,61,62,63,64,65]. The decomposition of ammonium sulfate occurs between 300 and 400 °C; thus, at lower temperatures, its accumulation significantly impairs catalyst denitration activity. For Mn-based oxide catalysts, SO_x leads to permanent deactivation through sulfate deposition, limiting their practical application. The deactivation process by SO₂ involves three stages: adsorption of SO₂, its oxidation to SO₃, and the subsequent deposition of ammonium sulfate or sulfation of active components. Wei et al. [66] highlighted that Mn-Ce/TiO₂ catalysts exhibit resistance to sulfur poisoning in low-temperature SCR, attributing deactivation to SO₂ outcompeting NH₃ for active sites, as evidenced by DRIFT and DFT analyses. Despite these insights, understanding the resistance mechanism against SO_x on Mn-based oxide catalysts remains challenging for low-temperature applications. Further research is essential to elucidate how SO_x suppresses de-NO_x activity at low temperatures, thereby enhancing the poison-resistant performance of Mn-based oxide catalysts.

Experimental studies on SCR have been limited by technical constraints, leaving critical questions about the reaction pathways and intermediates unresolved [7,8]. Therefore, an in-depth theoretical investigation employing DFT calculations is crucial for unveiling the fundamental aspects of the SCR process. In this work, DFT calculations of NH₃-SCR mechanism were carried out on the Mn₃O₄(001) surface. The adsorption configurations and adsorption sites of NH₃ and NO were determined. Through the calculated free energy changes in the reaction process, NH₃ dehydrogenation, the formation of important intermediates (NH₂NO and NHNO) and byproduct (N₂O) via N−N bonding, and the desorption of N₂ and H₂O were analyzed in detail. In addition, the adsorption of SO₂ and SO₃ on pure and NH_x pre-adsorbed Mn₃O₄(001) were discussed to explore the possible mechanisms of SO_x poisoning.

2. Results and Discussion

2.1. SCR Mechanism

2.1.1. Adsorption Properties of NH₃ and NO

Figure 1 shows the most stable adsorption configurations of NH₃ and NO on the Mn₃O₄(001) surface, and

Table 1. Adsorption sites, adsorption energies (ΔE_ads, eV), and bond lengths (d, Å) for adsorbed reactant molecules on Mn₃O₄(001).

Molecule	Site	Bond Length	ΔEads
NH3	atop-Mn	dN-Mn = 2.088	−0.971
NO	atop-Mn (O-Mn) a	dO-Mn = 2.115	−0.375
	bridge-Mn	dN-Mn = 1.758 dO-Mn = 2.241	−1.431
	atop-Mn (N-Mn) a	dN-Mn = 1.671	−1.864
SO2	atop-Mn	dS-Mn = 2.763	−0.284
	bridge-Mn	dO-Mn = 1.955, 1.949	−0.833
	hollow	dS-O = 1.662 dO-Mn = 1.983, 2.021	−1.451
SO3	bridge-Mn	dO-Mn = 1.842, 1.839	−1.424
	hollow	dS-O =1.597 dO-Mn = 2.005, 2.027	−2.301

^a “O-Mn” denotes that NO binds to the Mn atom via the O atom, while “N-Mn” refers to the N-Mn bonding.

lists the corresponding geometric parameters and adsorption energies. NH₃ could stably stay through the N atom at the top-Mn site with the N−Mn axis perpendicular to the surface. The N−Mn bond length is 2.088 Å, and the adsorption energy (ΔE_ads) is calculated to be −0.971 eV. Similar adsorption configurations were also obtained for NH₃ on other Mn-based oxide surfaces [41,67,68,69]. Yang et al. [41] proposed that the most stable adsorption site of NH₃ on Mn₃O₄(110) is the Mn site with the ΔE_ads of −1.00 eV. Chen et al. [40] also identified a top-Mn adsorbed NH₃ on Mn₃O₄(101), and the corresponding ΔE_ads is −0.94 eV. NO adsorption involves three geometric types:

(1) NO binds to the Mn atom via the O atom with the N−O axis vertical to the surface. The O−Mn bond length is 2.115 Å and the adsorption strength is weakly mirrored by the ΔE_ads of −0.375 eV. (2) When NO begins to adsorb in a horizontal orientation, a bridge-type adsorption configuration forms accordingly. In this configuration, the N and O atoms bond with two Mn atoms on the surface, and the corresponding bond lengths are 1.758 Å and 2.241 Å, respectively. The adsorption strength between NO and the surface increases accordingly, reflected by the lower ΔE_ads of −1.431eV. (3) NO can also adsorb vertically on the surface via the N−Mn bonding. The N−Mn distance of 1.671 Å is substantially shorter than above N−Mn and O−Mn bond lengths, indicating a stronger N−Mn interaction. This accounts for the lowest ΔE_ads value of −1.864 eV for NO binding at the top-Mn (N−Mn) site. On Mn₃O₄(101), for comparison, Chen et al. [40,70] confirmed a vertical adsorption configuration of NO via the N−Mn bond with the ΔE_ads of −2.37 eV. Zhang et al. [71] built a Mn₃O₄/TiO₂ catalyst model to study the adsorption properties of NO_x. They also identified the stable NO adsorption on the Mn site with the N−Mn bond length of 1.863 Å and the ΔE_ads of −2.66 eV.

2.1.2. Reaction Pathways

The SCR process can generally be divided into four stages: molecular adsorption and NH₃ dehydrogenation, N−N bonding, the formation of N₂ and H₂O, and surface regeneration [72]. The calculated free energy profiles for NH₃-SCR on Mn₃O₄(001) are displayed in Figure 2 with the corresponding intermediate structures presented in Figure 3. Two main reaction pathways are discussed in detail. One is the “E-R” mechanism, in which NH₃ adsorbs first, followed by dehydrogenation, and then the adsorbed NH₂ reacts with gas-phase NO. The other is the “L-H” mechanism, which starts with co-adsorbed NO and NH₃.

• The E-R Mechanism

The oxidative dehydrogenation of NH₃ to active NH₂ species was decisive for the E-R mechanism because NH₂ species could directly react with gaseous NO to generate important NH₂NO intermediate that subsequently decomposed to N₂ and H₂O. The NH₃ activation to V-NH₂ amide species over vanadyl centers has been demonstrated over bulk V₂O₅ [73], V₂O₅−TiO₂ [74], and V₂O₅−WO₃−TiO₂ [75] catalysts. Adsorbed NH₃* dissociates into NH₂* and H* species (Figure 3a) with an increase in free energy of 0.585 eV. Further step-wise dehydrogenations of NH₂* to produce NH* (Figure 3h) and N* (Figure 3i) are thermodynamically unfavorable due to the high endothermicity with the respective ΔG values of 0.620 and 1.609 eV. Following the classic E-R pathway [7,8], adsorbed NH₂* can react with a gas-phase NO molecule, forming a stable N–N bond with a bond length of 1.459Å. This step is endothermic, reflected by the ΔG value of 0.333 eV. The newly formed intermediate, NONH₂* (Figure 3b), still binds to the surface Mn site through the N atom of the NH₂ group. Then, NONH₂* converts to NH₂NO* (Figure 3c), in which the adsorbed configuration switches to the NO group bonding with the Mn site rather than NH₂. For NH₂NO*, the N−Mn bond length between the N atom of NO and Mn decreases to 2.152 Å compared with that of 2.181 Å between NH₂ and Mn for NONH₂*, indicating that this geometric transition from NONH₂* to NH₂NO* increases the interaction between adsorbed intermediate and surface, reflected by the exothermicity of this step (ΔG = −0.305 eV). The NH₂NO* intermediate further undergoes dehydrogenation by transferring one H atom to an adjacent surface O site, forming cis-NHNO* species (Figure 3o) and a second surface H* species. The free energy is reduced by 0.228 eV in this step. To facilitate the removal of the third H atom, cis-NHNO* converts to trans-NHNO* (Figure 3d) with the N−H axis pointing to the surface (ΔG = 0.238 eV). Accompanied by the cleavage of the N−H bond, a new N−Mn bond is formed, generating a π-bonding configuration of ONN* (Figure 3e) at the Mn site. This third dehydrogenation step is endothermic with a ΔG of 0.543 eV. Afterward, an adsorbed H₂O_L molecule (“L” represents a surface lattice O atom) can be produced by the combination of two H atoms and one surface lattice O_L atom (Figure 3f). This step requires the breaking of one O−H bond, which accounts for the high endothermicity with the ΔG of 0.959 eV. Subsequent H₂O desorption leaves one oxygen vacancy (O_V) on the surface (Figure 3g), followed by the refilling of the O_V by the O atom of the ONN* and the N₂ desorption. The formation of gas-phase H₂O and N₂ releases significant amounts of energy, as indicated by their respective ΔG values of −1.016 and −3.769 eV. These energy releases are primarily due to the refilling process of O_V by the O atom from the ONN*, which stabilizes the system and contributes to the overall exergonic nature of the reactions. After the reactions in the aforementioned three stages are completed, an adsorbed H atom remains atop one surface lattice O atom of Mn₃O₄(001) surface, forming a Brønsted acid site denoted as H* (Figure 3t). During the final stage of catalyst surface regeneration, the two residual H atoms combine with a lattice O atom from the surface to form H₂O, which then desorbs from the surface. The O_V created on the surface as a result of this process is subsequently replenished by 1/2 O₂.

For the H* Brønsted acid site formed at the end of the third stage (formation of H₂O and N₂), we further investigated if this H* site could act as an active center for the NH₃- SCR reaction. Generally, there are two kinds of acid sites on the surface of metal oxide catalysts, the Lewis acid site and Brønsted acid site. The adsorption of NH₃ can be divided into two categories according to the type of acid site: one is the NH₃ adsorbed on Brønsted acid sites, labeled as NH₄⁺; the other one is the NH₃ adsorbed on Lewis acid sites, labeled as NH₃*, which has been described above. Herein, the adsorption of NH₃ at the H* site to form ammonium species (NH₄*) was evaluated (Figure S1 in the Supplementary Materials). The results showed that the interaction between NH₃ and H* is characterized by a ΔE_ads value of only −0.559 eV, with no indication of N−H bond formation (d_N−H = 1.706 Å). This suggests that the binding of NH₃ at the H* site is weak, resulting in an unstable intermediate. Consequently, it is inferred that the Lewis acid sites (surface Mn sites), rather than the H* Brønsted acid sites, are the principal active centers for NH₃ adsorption. Moreover, the H* Brønsted acid site does not favorably interact with NO either. Therefore, the H* Brønsted acid site likely plays a passive role in the NH₃-SCR mechanism, acting more as a spectator than an active participant. In the final stage of the catalyst surface regeneration, the NH₃ and NO species adsorbed at the Mn-based Lewis acid sites proceed through the aforementioned three reaction stages twice, leading to the production of two surface H*. Then, adsorbed H₂O is formed through H transfer, and H₂O desorption leads to the formation of O_V. Finally, the O_V is replenished by 1/2 O₂, which regenerates the catalyst surface to the initial state and completes the NH₃-SCR catalytic cycle. The above results underscore the importance of the Lewis acid sites in facilitating the overall SCR process while highlighting the limited involvement of the Brønsted acid sites in the key reaction steps.

It is important to note that after the formation of the ONN* intermediate, besides the original E-R mechanism where H transfer occurs and results in the formation of H₂O_L, there is also a possibility for ONN* to directly desorb, producing gaseous N₂O (Figure 3j). In contrast to the highly endothermic pathway leading to H₂O formation, the N₂O desorption process is significantly exothermic, mirrored by the ΔG of −1.125 eV. Therefore, from a thermodynamic perspective, the production of N₂O is more favorable than the E-R pathway that produces H₂O followed by N₂. This is in good agreement with the DFT results on Mn₃O₄(110) by Yang et al. [41], in which the adsorbed NH₂NO*, generated by the reaction of adsorbed NH₂* and gaseous NO, tends to convert into N₂O rather than N₂. Our results show that on the Mn₃O₄ surface, the selectivity toward the byproduct N₂O is higher than that toward N₂, suggesting that the classic E-R mechanism for the NH₃-SCR of NO might be flawed and cannot fully account for the product distribution of the SCR process on Mn₃O₄ surface.

• The L-H Mechanism

The L-H pathway begins with the co-adsorption of NH₃ and NO (Figure 3k) on the Mn₃O₄(001) surface. NO and NH₃ adsorb on two neighboring Mn sites. Adsorbed NH₃ undergoes dehydrogenation, releasing one H atom (Figure 3l). This step is energetically favorable in the L-H mechanism (ΔG = 0.350 eV) compared to the E-R mechanism (ΔG = 0.585 eV), as indicated by the lower energy required for dehydrogenation. The presence of adsorbed NO alters the electronic environment of the surface oxygen atoms, stabilizing the adsorption of hydrogen and creating conditions that favor the further dehydrogenation of NH₂* to NH* (Figure 3m). Notice that the free energy changes in the N−N coupling between adsorbed NH₂* and NO* to form NONH₂* (1.203 eV) and NH₂NO* (0.898 eV) are higher than the energy required to dissociate the second hydrogen from NH₂* to form NH* + H* (0.637 eV), indicating that the reaction steps of NH₂*+ NO* → NONH₂*/NH₂NO* are thermodynamically unfavorable. The transition from NH₂* to NH* is still endothermic with the ΔG of 0.636 eV. Once formed, NH* interacts with NO* to produce the intermediate NONH* (Figure 3n). The intermediate NONH* is then transformed into cis-NHNO* (Figure 3o), which serves as a common intermediate where the E-R and L-H pathways converge. The next stage is the formation of H₂O and N₂. Adsorbed cis-NHNO* is initially isomerized to iso-NHNO* (Figure 3p), in which the O atom of the NO group binds to the Mn site instead of the N atom. This isomerization step is slightly exothermic with a ΔG of −0.127 eV. Then, one adsorbed H atom approaches a surface O_L−H group, forming an adsorbed H₂O_L (Figure 3q). Similar to the case in the E-R path, this H transfer process to generate H₂O is difficult, mirrored by the high ΔG value of 0.719 eV. Afterward, the O atom of iso-NHNO* occupies the position of O_L, producing adsorbed HNN*, while the original H₂O_L is lifted up over the surface and becomes adsorbed H₂O*, binding to an adjacent Mn site (Figure 3r). This step is slightly exothermic with a ΔG of −0.251 eV. The subsequent desorption of H₂O is also slightly exothermic (ΔG = −0.052 eV), but it does not involve the formation of an O_V (Figure 3s), which is different from the case in the E-R path (Figure 3g). After H₂O desorption, the N−N group within the HNN* intermediate subsequently desorbs from the surface in the form of N₂. Concurrently, the N−H bond breaks and the H atom is captured by a surface O atom, leading to the formation of a highly stable O−H bond (Figure 3t). The formation of this bond is substantially exothermic, resulting in a significant change in free energy (ΔG = −3.333 eV) for this N₂ desorption step. For the catalyst surface regeneration, the final stage of the L-H path merges with that of the E-R path.

In summary, the L−H mechanism is more favorable than the E−R mechanism because of the lower free energy profile. The key intermediates of the N−N bonding are the NH₂HO* in the E−R path and NHNO* in the L−H path, respectively, and the N−N bonding is easier for the L−H path thermodynamically. The potential-determining step (PDS) is the H transfer step for both the E−R and L−H paths, in which the cleavage of one O−H bond occurs accompanied by the formation of adsorbed H₂O_L. In the traditional L-H mechanism, gaseous NO is first oxidized to form adsorbed nitrites or nitrates, which then react with adsorbed NH_x species to ultimately produce H₂O and N₂. In contrast, a potential L-H pathway identified in this work involves gaseous NO first adsorbing and then reacting with NH* to form the intermediate NHNO*, which decomposes into H₂O and N₂. This new L-H pathway is more efficient as it bypasses the NO oxidation step and is more selective for N₂ formation by avoiding N₂O production.

2.2. SO_x Poisoning Mechanisms

2.2.1. Adsorption Properties of SO₂ and SO₃

SO₂: Three distinct adsorption configurations of SO₂ on the Mn₃O₄(001) surface were investigated (Figure 4). In the first configuration, the S atom of SO₂ is directly bonded to a surface Mn atom, forming an S−Mn bond with a length of 2.763 Å. This interaction results in a relatively weak ΔE_ads of −0.284 eV, indicating that this structure is less stable. The second configuration involves both O atoms of SO₂ being attracted by adjacent surface Mn atoms, leading to the formation of two O−Mn bonds with lengths of 1.955 Å and 1.949 Å, respectively. The strong interaction between the Mn and O atoms shortens these bond lengths compared to the first configuration, resulting in a more stable geometry with the ΔE_ads of −0.833 eV. The third and most stable configuration features SO₂ interacting with the Mn-O-Mn structure of the surface, forming one S−O bond and two O−Mn bonds. The corresponding bond lengths are 1.662 Å and 1.983/2.021 Å, respectively. This configuration exhibits the most negative ΔE_ads of −1.451 eV, signifying its stability and preference over the other two configurations. For comparison, Liu et al. [72] reported that the most stable adsorption of SO₂ on Ti-doped CeO₂ catalysts involves a direct S−Ti bond with a ΔE_ads of −3.010 eV, suggesting that Mn₃O₄ has a much weaker affinity for SO₂ than Ti-doped CeO₂ and thereby possesses certain anti-sulfurization properties. Xiong et al. [42] determined that adsorbed SO₂ forms two bonds with the Mn₃O₄(101) surface: one S−O bond with a surface lattice oxygen atom and one O−Mn bond with a surface Mn atom. Compared with the configuration on Mn₃O₄(001), where both oxygen atoms of SO₂ form bonds with surface Mn atoms, the SO₂ on Mn₃O₄(101) forms one fewer O−Mn bond but exhibits a stronger binding strength, as evidenced by its more negative ΔE_ads value of −1.79 eV. This indicates that Mn₃O₄ is facet-sensitive, and different crystal facets may exhibit different anti-poisoning performances. Compared with NO, which has an adsorption energy of −1.864 eV, SO₂ exhibits a lower adsorption strength on the Mn₃O₄(001) surface with an ΔE_ads value of −1.451 eV, indicating a small impact of SO₂ on NO adsorption.

SO₃: SO₃ exhibits two stable adsorption structures on the Mn₃O₄(001) surface (Figure 4). In the first upright structure, two S−O bonds of SO₃ form two O−Mn bonds with the surface, while the third S−O bond points away from it, with a ΔE_ads of −1.424 eV. The second flat structure evolves from the first, where the S atom additionally bonds with a nearby surface O atom, still with the third S−O bond slightly oriented away from the surface. This second structure has a more negative ΔE_ads of −2.301 eV, indicating that SO₃ adsorption is more competitive compared to NH₃, NO, and SO₂ on Mn₃O₄(001).

2.2.2. Formation of SO_x-NH and SO_x-NH₂ Complexes

Among the four stages of NH₃-SCR on Mn₃O₄(001), the N−N bond formation to create intermediates NH₂NO* (E-R path) and NHNO* (L-H path) is crucial. However, SO_x species in the reaction atmosphere compete with NO for adsorption sites on Mn₃O₄(001), and also for binding with NH₂* and NH*, potentially hindering N−N bonding and disrupting the NH₃-SCR process. Figure 5 illustrates the binding configurations of SO₂ and SO₃ with NH₂* and NH*. For the formation of NONH₂*, the adsorption energy of NO binding to the N atom of NH₂* is −0.515 eV. SO₂ binds more readily to NH₂* and occupies the N site with an adsorption energy of −1.342 eV, forming a more stable SO₂-NH₂* structure with a S−N bond length of 1.922 Å. For the formation of NONH*, the adsorption energy of NO binding to the N atom of NH* is −1.610 eV. SO₂ also binds to NH* to form SO₂-NH* with a S−N bond length of 1.701 Å and an adsorption energy of −1.536 eV. Consequently, while the formation of NONH* in the L-H path is less affected by SO₂ adsorption, the formation of NONH₂* in the E-R path is significantly impacted by SO₂ binding to NH₂*. Combining this with the ΔE_ads values of NO (−1.864 eV) and SO₂ (−1.451 eV) on Mn₃O₄(001), it can be concluded that the poisoning effects of SO₂ on the surface adsorption sites and the formation of key intermediates are limited, proving that the Mn₃O₄ surface has certain anti-SO₂ poisoning properties for the NH₃-SCR process. The adsorption energies of SO₃ binding to the N atom of NH₂* and NH* (Figure 5c,f) were calculated to be −2.304 and −2.803 eV, respectively, indicating a much higher binding strength compared with SO₂. Combining with the ΔE_ads of SO₃ (−2.301 eV) on Mn₃O₄(001), we found that the poisoning effects of SO₃ on the surface adsorption sites and the formation of key intermediates are substantially strong.

2.2.3. Formation of NH₄HSO₄

The conversion of SO₂ to SO₃ requires a reaction temperature above 400 °C. In contrast, the working temperature for Mn₃O₄, the low-temperature denitrification catalyst, is generally below 300 °C. Consequently, within the current catalytic system utilizing Mn₃O₄, SO₂ does not convert to SO₃, and thus it cannot further transform into NH₄HSO₄. The formation of NH₄HSO₄ primarily results from the existing SO₃. In the gas-phase mechanism for NH₄HSO₄ formation [10], SO₃ reacts with H₂O to form H₂SO₄, which then binds with NH₃ to produce NH₄HSO₄. For the most stable adsorption configuration of SO₃ on Mn₃O₄(001), however, the calculation of adsorbed SO₃ and gas-phase H₂O to form H₂SO₄ was not convergent. Instead, we found that vertically adsorbed SO₃ can sequentially react with gas-phase H₂O and NH₃ to yield NH₄HSO₄ (Figure 6), with corresponding free energy changes of −0.094 and −0.193 eV. For the newly formed adsorption structure of NH₄HSO₄, we observed that one O−H bond has broken, with the dissociated H atom relocating to an adjacent surface O atom, thereby creating a H* Brønsted acid site. The O−H distance is 2.010 Å, which falls within the typical range for a classic O−H hydrogen bond (1.5~2.2 Å), indicating a weak H−bond interaction. This H−bond helps stabilize NH₄HSO₄ on the Mn₃O₄ surface. The continuous exothermic nature of NH₄HSO₄ formation suggests thermodynamic feasibility and further poisoning of the Mn₃O₄ surface.

3. Computation Methods

DFT calculations were performed with the Vienna ab initio simulation package (VASP) [76,77] using the projector augmented wave method (PAW) [43,78] to describe the interaction between core and valence electrons. The exchange correlation effect was processed using generalized gradient approximation developed by Perdew–Burke–Ernzerhof (GGA-PBE) [79]. Plane waves were included for the electronic wave functions up to a cutoff energy of 400 eV. The convergence criteria for the energy and force were set at 10⁻⁵ eV and −0.03 eV/Å, respectively. The Gaussian smearing method with a width of 0.05 eV was used to speed up the convergence of electronic structure optimization. All calculations were performed with spin polarization. The lattice constants of trimanganese tetroxide (Mn₃O₄) were fitted by the Birch–Murnaghan equation of state, and the Brillouin zones were sampled with 7 × 7 × 7 k-point meshes [80]. The lattice constant of Mn₃O₄ was calculated to be 6.286 Å, in good agreement with the experimental value (6.345 Å) [67]. This calculated lattice constant was subsequently used in all calculations.

As shown in Figure 7, the Mn₃O₄(001) surface was built with a 2 × 2 periodic slab model, consisting of nine atomic layers and a vacuum gap of 15 Å. Full geometry optimizations were performed for all relevant adsorbates and the uppermost four atomic layers of the Mn₃O₄(001) slab without symmetry restriction, while the bottom five atomic layers were fixed in the positions at the calculated lattice constant. The reciprocal space was sampled by a grid of (3 × 3 × 1) k-points generated automatically using the Monkhorst–Pack method [80]. The (001) surface was chosen because it is the most favored and involves the lowest cleavage energy (2.33 J·m⁻²) and surface energy (1.40 J·m⁻²) compared with other crystallographic planes, such as (100), (101), and (110) [81]. The (001) surface consists of two alternating terminations, denoted as Mn_tet and Mn_oct-O. The Mn_tet termination is composed of tetrahedral Mn²⁺ cations. The Mn_oct-O termination comprises Mn³⁺ with octahedral coordination and O^2- anions. The Mn_tet layer terminated (001) surface is not stable and will undergo reconstruction, similar to the case on the Fe-rich Fe₃O₄(001) surface, while the Mn_oct-O layer terminated (001) surface remains stable and exhibits high catalytic activity [82,83]. Therefore, the Mn_oct-O termination was adopted in this work.

The adsorption energy (ΔE_ads) was calculated using the following formula:

ΔE_ads = E_{adsorbate/substrate} − E_adsorbate − E_substrate

(1)

In the formula, E_adsorbate and E_substrate are the total energies of the free adsorbate molecule and the substrate, respectively. The E_{adsorbate/substrate} is the total energy of the adsorbate molecule adsorbed on the substrate. In this definition, the negative value of ΔE_ads corresponds to the stable adsorption.

Without considering the contribution of thermal internal energy, the Gibbs free energy change of each elementary step is determined as

ΔG = ΔE + ΔE_ZPE −TΔS

(2)

where ΔE is the reaction energy of each elementary step and ΔE_ZPE and ΔS are the zero point energy (ZPE) difference and the reaction entropy difference between the product and reactant of each elementary step, respectively. In this work, the vibrational entropies were considered and the vibrational frequencies (ν) were calculated by the finite difference method [68]. The ZPE of each adsorbed intermediate can be expressed as $E_{\mathrm{ZPE}}={\sum }_i\frac{1}{2}h{v}_i$, and the vibrational entropy contribution (S_νib) was calculated by the following standard equation from statistical mechanics [69]:

$$S_{\mathrm{vib}}=R\sum _i(\frac{\frac{{\mathsf{\Theta }}_{\mathit{vi}}}{\mathrm{T}}}{\exp \left(\frac{{\mathsf{\Theta }}_{\mathit{vi}}}{T}\right)-1}-\ln [1-\exp \left(\frac{-{\mathsf{\Theta }}_{\mathit{vi}}}{T}\right)\left]\right)$$

(3)

where R is the gas constant, T is the absolute temperature (500 K), and Θ_νi is the characteristic vibrational temperature, which equals hν_i/k_B. Here, h is Planck’s constant, ν_i is the vibrational frequency, and k_B is Boltzmann’s constant. For each gas phase species, the entropy contribution is obtained from NIST database [84]. This method has been successfully applied to describe redox reactions of nitrogen and oxygen successfully [85].

4. Conclusions

DFT calculations were performed to study the intrinsic mechanisms of NH₃-SCR and SO_x poisoning on Mn₃O₄(001). Both NH₃ and NO adsorb atop the surface Mn site via N−Mn bonding, and the corresponding adsorption energies were calculated to be −0.971 and −1.864 eV, respectively. The Lewis acid site (surface Mn site) is the principal active center for NH₃ and NO adsorption, while the H* Brønsted acid site does not favorably interact with either NH₃ or NO, demonstrating the importance of the Lewis acid sites in facilitating the overall SCR process, while highlighting the limited involvement of the Brønsted acid sites in the key reaction steps. The L-H mechanism is more favorable than the E-R mechanism because of the lower free energy profile. The key intermediates of the N−N bonding are the NH₂HO* in the E−R path and NHNO* in the L−H path, and the N−N bonding is easier for the L−H path thermodynamically. The potential-determining step is the surface H transfer step for both the E−R and L−H paths, in which the cleavage of one O−H bond occurs accompanied by the formation of adsorbed H₂O_L. In contrast to the traditional L-H mechanism in which gaseous NO is first oxidized to form adsorbed nitrites or nitrates and then react with adsorbed NH_x species to produce H₂O and N₂, a potential L-H pathway is identified in this work, which involves gaseous NO first adsorbing and then reacting with NH* to generate the key intermediate NHNO*, followed by the formation of H₂O and N₂. This new L-H pathway is more efficient as it bypasses the NO oxidation step and is more selective for N₂ formation by avoiding N₂O production.

Based on the exploration of the surface NH₃-SCR process, the poisoning mechanisms of SO_x on Mn₃O₄(001) are initially revealed. SO_x species in the reaction atmosphere compete with NO for adsorption sites on Mn₃O₄(001) and also for binding with NH₂* and NH*, potentially hindering N−N bonding and disrupting the NH₃-SCR process. Compared with NO, SO₂ exhibits a lower adsorption strength on Mn₃O₄(001) with the adsorption energy of −1.451 eV, indicating a small impact of SO₂ on NO adsorption. For the formation of NONH₂*, the adsorption energy of NO binding to NH₂* is only −0.515 eV, while SO₂ binds more readily to NH₂* and occupies the N site with an adsorption energy of −1.342 eV, forming a more stable SO₂-NH₂* structure. For the formation of NONH*, the adsorption energy of NO binding to NH* is −1.610 eV, while SO₂ can also bind to NH*, forming SO₂-NH* with an adsorption energy of −1.536 eV. Consequently, the formation of NONH₂* in the E-R path can be significantly impacted by SO₂ binding to NH₂*, while the formation of NONH* in the L-H path is less affected by SO₂ adsorption to NH*. The poisoning effects of SO₂ on the surface adsorption sites and the formation of key intermediates are limited, proving that the Mn₃O₄ surface has certain anti-SO₂ poisoning properties for the NH₃-SCR process. With a more negative adsorption energy of −2.301 eV, SO₃ adsorption is more competitive compared to NH₃, NO, and SO₂ on Mn₃O₄(001). The adsorption energies of SO₃ binding to the N atom of NH₂* and NH* were calculated to be −2.304 and −2.803 eV, respectively, indicating a much higher binding strength compared with SO₂. In addition, NH₄HSO₄, as a poison to the Mn₃O₄ surface, can be generated from SO₃ via the reaction with NH₃ and H₂O. Compared with SO₂, SO₃ exhibits substantially stronger poisoning effects, which include occupying adsorption sites, hindering the formation of key intermediates, and leading to the formation of ammonium bisulfate. This study provides detailed insights into the NH₃-SCR reaction mechanisms, as well as the SO_x poisoning mechanisms, thereby guiding the rational design of more efficient and durable Mn-based SCR catalysts.