Mechanistic insight into the selective catalytic reduction of NO by NH₃ over α-Fe₂O₃ (001): a density functional theory study

Abstract

The adsorption properties and the selective catalytic reduction mechanism of NO, NH₃ and O₂ molecules over the α-Fe₂O₃ (001) surface were studied by density functional theory. The SCR reaction on the α-Fe₂O₃ (001) surface mainly followed the Eley–Rideal-type mechanism and three detailed reaction routes were proposed. The reaction of NO and NH₃ consists of seven consecutive steps: (1) NH₃ adsorption, (2) NH₃ oxidation, (3) NH₂NO formation, (4) NH₂NO oxidation, (5) ONNH rearrangement, (2) ONNH dissociation and (7) H₂O formation. –NHNO undergoes rearrangement before it dissociates to form a N₂ molecule. NH₃ dissociation and H₂O generation are the rate-determining steps. NO oxidation has two reaction paths, NO₂ generation and –NO₃ formation. The reaction of NH₃, NO and O₂ mainly consists of three consecutive steps: (1) –NO₂NH₂ formation, (2) –NO₂NH formation, and (3) –NO₂NH dissociation. The combination of free NH₂ and –NO₂ forms a –NH₂NO₂ group, which releases an energy of 288 kJ mol⁻¹ and plays an important role in the overall reaction as it provides energy in the following reactions. The formation and dissociation of H₂O determine the SCR deNO_x reaction rate and the adsorbed oxygen promotes the SCR deNO_x reaction on the α-Fe₂O₃ (001) surface.

---

1. Introduction

Nitrogen oxides (NO_x) generated during the combustion of fossil fuels are the main sources of air pollution such as acid rain, smog, and photochemical smog. About 95% of NO_x emissions in the atmosphere come from automobile exhaust and coal-fired power plants.^1,2 At present, most power plants adopt the selective catalytic reduction (SCR) of NO_x with NH₃ in which a V₂O₅/TiO₂ catalyst is promoted by WO₃ or MoO₃.³ Although the V₂O₅/TiO₂–WO₃ or V₂O₅/TiO₂–MoO₃ catalyst has high deNO_x efficiency, quite a few problems still exist; for example, they easily oxidize SO₂ to SO₃ and form N₂O at high temperature, and the post-processing of the deactivated catalyst is also a challenge owing to the toxicity of vanadium. Therefore, it is imperative to discover and develop nontoxic and efficient catalysts for the reduction of NO_x.

Fe-based catalysts have shown good deNO_x performance. As early as the 1980s, Bosch and Janssen⁴ have tested the feasibility of unsupported Fe₂O₃ and several kinds of iron ore in the catalytic oxidation of NO_x. Ramis et al.⁵ studied the transformation and interaction of NO, NH₃, N₂H₄ and NH₂OH on Fe₂O₃/TiO₂ catalysts and found that the lattice structure of Fe₂O₃ had a great influence on the adsorption mode of molecules and deNO_x efficiency. Zhao et al.⁶ prepared different crystal Fe₂O₃ catalysts by a co-precipitation method and investigated their deNO_x properties at 250–450 °C. The results showed that the crystal shape of Fe₂O₃ had a marked effect on the deNO_x performance and the α-Fe₂O₃ catalyst had excellent NH₃-SCR activity. In recent years, the deNO_x performance of Fe₂O₃-based catalysts has been further enhanced with the improvement of the preparation method. Han et al.⁷ prepared carbon nanotube-supported Fe₂O₃ nanoparticles (Fe₂O₃/CNTs) and found that this catalyst had an excellent deNO_x performance owing to its high dispersibility, more powerful reducibility, strong acidity and high chemical adsorption oxygen content. They synthesized Fe₂O₃/CeO₂ nanorods and also obtained higher NO conversion.⁸ A H₃PW₁₂O₄₀·6H₂O-modified ring-like Fe₂O₃ catalyst exhibited not only excellent NH₃-SCR activity at 250–500 °C but also outstanding resistance against SO₂ and H₂O.⁹ Moreover, the α-Fe₂O₃ catalyst has higher strength and better heat resistance than the γ-Fe₂O₃ catalyst. Hence, the α-Fe₂O₃ catalyst has great potential as an nontoxic SCR deNO_x catalyst.

Density functional theory (DFT) has been proved to be a good method to study the SCR of NO_x with NH₃. Liu et al.¹⁰ investigated the facet–activity relationship of Fe₂O₃/TiO₂ nanocatalysts in the NH₃-SCR reaction using the DFT method. Maitarad et al.¹¹ confirmed the possibility of using metal–purine catalysts for direct decomposition of N₂O by DFT calculations. Yan et al.¹² demonstrated that the adsorption behavior of NH₃ and NO on the surface of the 0.5 μm MnO_x–FeO_y nanocage was different from that of other scales by DFT calculations and experimental studies. Esrafili¹³ studied the reduction of NO molecules on Si-doped boron nitride nanosheets using dispersion-corrected DFT calculations. Many previous studies have been devoted to examining the activity of the Fe₂O₃ catalyst by experiment and theoretical calculation.^14–19 However, the mechanism of NO_x removal from flue gas using the Fe₂O₃ catalyst as a SCR catalyst has been ignored. Although studies of the SCR deNO_x mechanism using density functional theory (DFT) have achieved excellent results,^20–22 most of them focused on commercial catalysts based on titanium,^23–30 and few studies concentrated on the reaction route of nontoxic SCR catalysts, especially Fe-based catalysts. In this paper, DFT was introduced to study the adsorption properties of NO, NH₃, and O₂ molecules over the α-Fe₂O₃ catalyst, and then the potential energy surface (PES) of formation and decomposition of complexes was systematically studied to investigate the reaction route and the rate-determining step (RDS) of the SCR of NO_x with NH₃ for the α-Fe₂O₃ catalyst.

2. Computational methods

DFT calculations were carried out with the Vienna Ab initio Simulation Package (VASP) version 5.3.^31–33 For exchange-correlation functional, the projector augmented wave (PAW) method and the Perdew–Burke–Ernzerhof (PBE)³⁴ functional were used for all calculations. The electron wave function is expanded by a plane wave whose kinetic energy cutoff is 500 eV. The DFT+U method is used to solve the strong electron correlation in atoms of transition metals and their oxide systems.^35–37 Hence, in this paper, a value of U = 4.0 eV was used on the d orbital of Fe, which is in excellent agreement with experimental data.³⁸ The atomic force and convergence accuracy of energy values were set as 0.03 eV A⁻¹ and 1 × 10⁻⁴ a.u., respectively. As shown in Fig. 1(a), the crystal structure of α-Fe₂O₃ has hexagonal close packing with the space group R3̄c; the lattice parameters are a = b = 5.035 Å, c = 13.74 Å and the distance between adjacent Fe–O atoms is 1.960 Å.³⁹ For the α-Fe₂O₃ bulk calculations, the energies are converged with (5 × 5 × 3) K points in a Monkhorst–Pack grid. The optimized unit cell parameters (a = b = 5.065 Å, c = 13.843 Å) are within +0.60% and +0.75% error compared with the experimental values, indicating that the calculation is reliable.

Fig. 1 Slab models of the α-Fe₂O₃ (001) surface. (a) α-Fe₂O₃ unit cell, (b) side view of the α-Fe₂O₃ (001) surface, and (c) top view of the α-Fe₂O₃ (001) surface with two different adsorption sites.

Weiss and Ranke found that the α-Fe₂O₃ (001) surface is catalytically active.⁴⁰ Moreover, the Fe–O₃–Fe termination of the (001) surface was proved to be the most stable among a series of terminations.^39,41,42 Hence, the α-Fe₂O₃ (001) surface was chosen to calculate the adsorption energies and reactive routes of NH₃, NO and O₂. A periodic slab with a 2 × 2 surface unit cell was chosen and spin-polarized calculations were used in all DFT computations.

According to the twelve-layer periodic slab model that we used, the top six surface layers and the adsorbed gas molecules were completely relaxed, and the six bottom layers were fixed (Fig. 1b). As shown in Fig. 1c, two different adsorption sites (Fe-top and Os) were distributed on the surface. A 15 Å vacuum gap in the z-direction was introduced to simulate the surfaces. A Monkhorst pack 1 × 1 × 1 mesh was used for the Brillouin zone integration because of the large size of the slab model (10.13 Å × 10.13 Å × 25.12 Å).²¹ Because adsorption was allowed on only one side of the exposed surfaces, the accordingly corrected dipole moment was introduced in the z–direction.⁴³ The gas molecules (NH₃, NO and O₂) can be optimized by putting them in a crystal lattice of 10 Å. The total adsorption energy of gas molecules on the α-Fe₂O₃ (001) surface (E_abs) was calculated using eqn (1):

E_abs = E_{gas+α-Fe2O3 (001) surface} − (E_gas + E_{α-Fe2O3 (001) surface}) (1)

where E_{gas+α-Fe2O3 (001) surface} is the total energy of the adsorbed system, E_{α-Fe2O3 (001) surface} is the total energy of the α-Fe₂O₃ (001) surface, and E_gas is the energy of the isolated gas molecule (NH₃, NO and O₂) in a vacuum. To search for a minimum energy path (MEP) and get the exact configuration of the reaction transition state, the climbing image-nudged elastic band (CI-NEB) method developed by the Henkelman group was introduced in this paper.⁴⁴ Its advantage is that when the configuration near the saddle point is not affected by the spring force, it freely relaxes to the position of the exact transition state so as to obtain the most real total transition state energy. This method not only can search for MEP, but also can accurately get the transition state of the reaction and the accurate reaction energy barrier.

3. Results and discussion

3.1 Adsorption of NH₃, NO and O₂ on the α-Fe₂O₃ (001) surface

3.1.1 Single-molecule adsorption. It is well established that the Langmuir–Hinshelwood (L–H) and Eley–Rideal (E–R) mechanisms were widely applied in the research of the NH₃-SCR reaction. The L–H mechanism is that in which free NO_x and NH₃ molecules are chemically adsorbed on the surface of the catalyst and then react with the adsorbed –NO_x and –NH₃ species, while in the E–R mechanism NH₃ or NO_x molecules react with the chemically adsorbed molecules on the catalyst surface. Whether the E–R or the L–H mechanism is used, the adsorption of molecules will occur over the catalyst surface. Nevertheless, the adsorption performance of molecules on the surface of the α-Fe₂O₃ catalyst is not yet clear; hence the adsorption of NO, O₂ and NH₃ over the α-Fe₂O₃ (001) surface are first studied.

The gas molecules are placed about 3.5 Å above the α-Fe₂O₃ (001) surface and then their structures are optimized. The optimized configurations of single molecules adsorbed on the α-Fe₂O₃ (001) surface are shown in Fig. 2. The gas molecules of NO, O₂ and NH₃ tend to adsorb on the Fe-top site and are not sensitive to other adsorption sites. Compared with the previous position (3.5 Å), the distance between the gas molecule and the catalyst surface becomes shorter, and the distance between the atoms within the molecule also changes accordingly. The bond lengths of NH₃, NO and O₂ molecules increase from 1.021 to 1.023 Å, 1.169 to 1.180 Å and 1.23 to 1.266 Å, respectively. The change of bond length of NH₃ is small, demonstrating that the N–H bond is difficult to activate, while the change of O–O bond length is large, indicating that the O₂ molecule is activated and participates in the reaction. To further understand the bond characteristics and charge transfer of NO, O₂ and NH₃ on the α-Fe₂O₃ (001) surface, the adsorption energy, Bader charge change and projected density of states (PDOS) of the single-molecule adsorption were calculated. The results are shown in Table 1 and Fig. 3.

Fig. 2 Configurations of the single molecule adsorbed on the α-Fe₂O₃ (001) surface: (a) NH₃, (b) NO, and (c) O₂. O: red, Mn: gray, H: white, and N: blue.

Table 1 Adsorption energies and Bader charge changes for NH₃, NO, and O₂ molecules on the surface of α-Fe₂O₃ (001)

Gas molecule	E ads/(kJ mol^−1)	Bader charge changes
a The Fe atom which bonds with gas molecules.
NH3	−115	N: +0.20, 3 × H: −0.06, Fe:^a −0.03
NO	−104	N: +0.26, O: −0.21, Fe: +0.2
O2	−123	^1O: +0.18, ^2O: +0.02, Fe: −0.03

---

Fig. 3 Projected density of states (PDOS) of NH₃, NO and O₂ molecules adsorbed on the α-Fe₂O₃ (001) surface.

As can be seen from Table 1, the adsorption energies of NH₃, NO and O₂ are −115, −104, and −123 kJ mol⁻¹, respectively, indicating that all of them are chemical adsorption energies. The amount of charge transfer of the three gas molecules can be obtained according to the Bader charges of NO, NH₃, and O₂ molecules. For the NH₃ molecule, the total charge increases after adsorption and the electrons in the N atoms transfer to Fe and H atoms, indicating that the bond between N and Fe is strengthened. Charge variation of the NO molecule shows that electrons transfer from Fe to N atoms and some of the electrons transfer from N to O atoms after adsorption, which leads to a strengthening of the N–O bond, implying that the adsorbed NO molecule is not conducive to the activation of the next step. The charges of the ¹O and ²O atoms increase, while the increase in charge of the O₂ molecule is obviously different from the decrease in charge of the Fe atom, suggesting that the increased charges come from the other atoms of the catalyst. PDOS shows that the density peaks for p orbitals of the N atom and d orbitals of the Fe atom of the α-Fe₂O₃ (001) surface are in resonance (shown in Fig. 3), indicating that orbital hybridization occurs between N and Fe atoms after NH₃ adsorption. As for the NO and O₂ molecules, some peaks of p orbitals of the N atom (NO) as well as the O atom (O₂) and d orbitals of the Fe atom overlap due to orbital hybridization. From these results, it can be concluded that NO, O₂ and NH₃ molecules adsorb on the Fe-top site with strong chemical adsorption.

3.1.2 Double-molecule adsorption. Two kinds of gas molecules may be adsorbed at the same adsorption site when gas molecules such as O₂, NO, and NH₃ are adsorbed on the α-Fe₂O₃ (001) surface. In the case of a common adsorption position, a precursor of subsequent deNO_x may be produced. To further study the adsorption behaviour of O₂, NH₃, and NO molecules on the α-Fe₂O₃ (001) surface, the double-molecule adsorption performances were studied. Configurations of double molecules adsorbed on the α-Fe₂O₃ (001) surface are shown in Fig. 4. O₂, NH₃, and NO are adsorbed on the Fe-top, and gas molecules cannot simultaneously adsorb at the same site to form chemical bonds. The adsorption energy and Bader charge change of double-molecule adsorption were calculated to gain more insight into the adsorption performance. The results are shown in Table S1.† There are some differences of the adsorption energy between single molecules and double molecules. The adsorption energy of two molecules for O₂ involved in the reaction such as NO + O₂ and NH₃ + O₂ is higher than that of the sum of a single molecule (shown in Table 1), suggesting that the O₂ molecule promotes the adsorption performance of the other gas molecule; that is, the O₂ molecule and the NO or NH₃ molecule have a synergistic effect. In addition, it can also be found from Table S1† that the adsorption energy of NO + O₂ is the largest, indicating that the adsorption of NO and O₂ can release large amounts of energy, which is beneficial for the subsequent reaction. From the Bader charge analysis of double-molecule absorption, it can be concluded that the NH₃ molecule cannot get more electrons compared with a single NH₃ molecular adsorption, suggesting that the electron-accepting ability of the NH₃ molecule is poorer than that of NO + O₂, which may be due to the low electronegativity of the N atom in the NH₃ molecule. The poor electron-accepting performance of the NH₃ molecule in the bimolecular adsorption indicates that its activity is low in the subsequent reaction. PDOS (shown in Fig. S1†) shows that there is no orbital hybridization between the gas molecules, indicating that it is difficult for the adsorbed gas molecules to react because of the greater distance, and thus the SCR deNO_x reaction on the α-Fe₂O₃ (001) surface mainly follows the E–R mechanism. Liu et al.⁴⁵ experimentally studied the SCR deNO_x performance of α-Fe₂O₃ and γ-Fe₂O₃ and found that the adsorbed NH₂ on the surface of α-Fe₂O₃ can react with a gas-phase NO molecule, indicating that the reaction path follows the E–R mechanism on the surface of α-Fe₂O₃, which is consistent with our theoretical calculation results. Therefore, the subsequent study of the reaction path is mainly focused on the reaction between adsorbed molecules and gas-phase molecules.

Fig. 4 Configurations of double molecules adsorbed on the α-Fe₂O₃ (001) surface. O: red, Mn: gray, H: white, and N: blue.

3.2 Reaction mechanism of NH₃-SCR

Although the basic reaction mechanisms of NH₃-SCR of NO have been studied,^21,46–49 different catalysts have different reaction mechanisms, especially for the α-Fe₂O₃ catalyst, for which the deNO_x mechanism is still unclear. Our proposed reaction mechanism for NH₃-SCR of NO over α-Fe₂O₃ (001) comprises three elementary schemes: Scheme 1, reaction of NO and NH₃; Scheme 2, NO oxidation; and Scheme 3, reaction of NH₃, NO and O₂.

Scheme 1 Reaction of NO and NH₃

NH₃ adsorption → NH₃ oxidation → NH₂NO formation → NH₂NO oxidation → ONNH rearrangement → ONNH dissociation → H₂O formation

Scheme 2 NO oxidation

Path 1: NO₂ generation and adsorption

Path 2: –NO₃ formation

Scheme 3 Reaction of NH₃, NO and O₂

–NO₂NH₂ formation → –NO₂NH formation → –NO₂NH dissociation.

3.2.1 Reaction of NO and NH₃. The CI-NEB results of the reaction of NO and NH₃ are shown in Fig. 5. NH₃ is a reducing agent of the SCR reaction, and thus its oxidation is a very important step. The NH₃ molecule adsorbs on the Fe-top site of the α-Fe₂O₃ (001) surface to form the Fe–NH₃ adduct, whose calculated adsorption energy is −115 kJ mol⁻¹ and intermolecular distance is 2.141 Å. In the transition state TS1′, a hydrogen atom from the adsorbed NH₃ approaches an oxygen atom to a distance of 1.163 Å. In the final product of this step, one H atom of the NH₃ molecule dissociates and binds to an O atom on the surface to form a hydroxyl species and the Fe–NH₂ adduct has a bond distance of 1.833 Å (state III′). The oxidation of the NH₃ molecule has an activation energy of 85 kJ mol⁻¹, and the energy of the final state in this reaction is higher than that of the initial state of 46 kJ mol⁻¹, suggesting that the first step in NH₃ dehydrogenation needs to consume more energy, which is consistent with the above adsorption performance analyses. Moreover, Li et al.,⁵⁰ Liu et al.⁵¹ and Zhang et al.⁵² detected –NH₂ species by in situ DRIFTs experiments. NH₂ dissociation into NH and the formation of the second hydroxyl were further studied. The high barrier energy calculated for the endothermic step causes the difficulty of N–H bond dissociation, indicating that further dissociation of NH₂ cannot take place. Song et al.²¹ studied the deNO_x mechanism of the MnCe_1−xO₂ (111) surface and also found that the dissociation of NH₂ is highly improbable.

Fig. 5 Energy and configuration profiles for the reaction route in Scheme 1. O: red, Mn: gray, H: white, and N: blue.

During the NH₃-SCR of NO, the Fe–N bond of FeH–NH₂ is broken and the intermediate interacts with a gaseous NO molecule to form the key intermediate, FeH–NH₂NO (state IV′). In this step, the Fe–N bond breakage has an activation energy of 11 kJ mol⁻¹, implying that the formation of this intermediate was relatively undemanding. In the transition state TS2′, the distances between N in NH₂ species and Fe and N of NO were 2.598 and 1.820 Å, respectively. NO adsorbs on FeH–NH₂ and a new N–N bond forms in the H₂N–NO intermediate with an N–N bond distance of 1.332 Å. In addition, it can be found by observing the geometric shape of the configuration that the structure of –NONH₂ is a planar structure, that is, the five atoms are on the same plane, and this reaction releases 101 kJ mol⁻¹ of energy with a minuscule energy barrier, indicating that the reaction easily occurs. In the transition state TS3′, a hydrogen atom from the adsorbed NH₂NO approaches an oxygen atom to a distance of 1.871 Å. In the final product of the second dehydrogenation, one H atom of NH₂NO dissociates and binds to the O atom on the surface to form the second hydroxyl species, and the Fe–ONNH¹ adduct has a N–N bond distance of 1.288 Å (state V′). The NH₂NO oxidation has an activation energy of 44 kJ mol⁻¹. We carried out rotation change and structural optimization on the –ONNH¹ structure because of its poor stability and found that the energy of structure VI′ is the lowest, −143 kJ mol⁻¹. In the NHNO rearrangement, there is no energy barrier for the change from structural V′ to ONNH² and the reaction is exothermic (ΔE = 15 kJ mol⁻¹). In a previous study, the NH₂NO intermediate rearranged to NHNOH before its decomposition has been found.⁵³ However, the NHNO rearrangement is absent. In the transition state TS4′, a hydrogen atom from the adsorbed ONNH² approaches an oxygen atom to a distance of 2.214 Å. In the final product of –ONNH² decomposition to N₂, H and O atoms connect to generate OH after –NONH (state V′) rotated and N₂ formed (state VII′). N₂ molecule generation and the subsequent desorption reaction energy barrier is smaller, which is 29 kJ mol⁻¹. The formation of N₂ is an exothermic process, which releases a large amount of energy (ΔE = 277 kJ mol⁻¹), suggesting that the reaction is a relatively stable process and the product is not easily converted back to the reactants.

In the following step, the bonds between H atoms and lattice oxygen on the surface of α-Fe₂O₃ (001) are broken and H bonds with –OH groups to form –H₂O species after N₂ desorption (state VIII′). This step is an endothermic reaction (ΔE = 130 kJ mol⁻¹). A large amount of energy generated by N₂ generation and dissociation can provide the energy to activate O–H bonds and generate H₂O. According to the path and the energy variation of the reaction of NH₃ with NO, it can be found that the active energy of NH₃ dissociation and the required energy of H₂O formation are relatively high; hence the two reactions are the rate-determining step (RDS) for the reaction of NH₃ with NO (Scheme 1).

3.2.2 NO oxidation. The presence of the O₂ molecule can generate more oxidation surfaces and promote the adsorption of NO.^54,55 According to the above studies on the adsorption properties of O₂ and NO, it can be found that they do not easily react after adsorption. Therefore, we tested the possibility of the reaction between gas molecules in the gas phase and adsorbed gases and found that the O₂ molecule does not easily react with adsorbed NO molecule, while NO easily reacts with adsorbed O₂, so we further studied the reaction mechanism of NO with adsorbed O₂. The reaction path and energy change (Scheme 2) are shown in Fig. 6.

Fig. 6 Energy and configuration profiles for the reaction route in Scheme 2. O: red, Mn: gray and N: blue.

The N atom of the NO molecule can naturally bind to the top O atom on Fe–O₂ to form Fe–O₂NO, with Fe–N and O–N bond lengths of 1.972 and 1.657 Å, respectively (state II′′). The released energy of the NO molecule adsorbing on Fe–O₂ is 85 kJ mol⁻¹, suggesting that this step is a facile reaction. In the following step, the long strip structure of Fe–O₂NO may be dissociated to produce a NO₂ molecule (path 1) or may be transformed to Fe–NO₃ (path 2). The Fe–NO₃ configuration, known as bidentate coordination nitrate, is one of the nitrogen oxides that is difficult to eliminate in nitrates. Although it has been detected during the experiment,⁵⁶ it was not found in a previous theoretical study. It can be seen from path 1 in Fig. 6 that the Fe–O₂NO group absorbs energy (ΔE = 45 kJ mol⁻¹) and the O–O bond breaks to form the NO₂ molecule. In the following step, the NO₂ molecule adsorbs at an adjacent Fe-top site to form a half-bridge nitrite (state IV′′), which has an energy barrier of 30 kJ mol⁻¹ and is associated with energy release (ΔE = 52 kJ mol⁻¹). In the transition state TS2′′, an oxygen atom from the NO₂ molecule approaches an adjacent iron atom to a distance of 2.307 Å.

For path 2, the transformation from Fe–O₂NO (state II′′) to Fe–NO₃ (state V′′) had an energy barrier of 122 kJ mol⁻¹ and released a large amount of heat (ΔE = 166 kJ mol⁻¹). From the comparison of the energy of structure II′′ and that of structure V′′, it can be found that Fe–NO₃ is more stable. As to the transition state TS1′′, the O–O bond breaks and an O atom approaches the Fe atom to a distance of 3.063 Å. Arnarson et al.⁵⁷ proposed a complete reaction mechanism for standard and fast NH₃-SCR on the low coverage VO_x/TiO₂ (001) surface and found that the energy barrier of NO reacting with adsorbed O₂ to form NO₂ was about −123 kJ mol, indicating that the generation of NO₂ requires relatively more energy compared to that of the α-Fe₂O₃ (001) surface. This may be due to the different adsorption behaviors of the O₂ molecule. According to the reaction paths and the energy changes of NO and O₂, it can be seen that the reaction of NO with O₂ to form NO₂ is easier than the formation of Fe–NO₃. Zhang et al.⁵⁸ studied the adsorption of related small gases for the NH₃-SCR reaction on the Mn-MOF-74 surface and also found NO₂ formation via NO oxidation with little barrier. In addition, bidentate ligand nitrates have good stability and can cover the catalyst surface in the SCR deNO_x reaction and decrease the efficiency of NO_x removal, so it is important to note that the formation of bidentate ligand nitrates should be avoided to the utmost extent in the SCR deNO_x reaction. In the NO oxidation reaction (Scheme 2), the O₂ molecule is adsorbed vertically on the catalyst surface before Fe–O₂NO formation. However, in our recent theoretical study of β-MnO₂,³⁷ the O₂ molecule was adsorbed horizontally on the β-MnO₂ surface and then bidentate ligand nitrates were directly generated when NO was added, so we speculate that the adsorption patterns of the O₂ molecule will have, to a certain extent, an influence on the formation of bidentate ligand nitrates. Li et al.⁵⁰ studied the adsorption of NO + O₂ on Fe₂O₃ using in situ DRIFTs and found the bands of linear nitrites, monodentate nitrite and nitro compound, which was consistent with our DFT calculations.

3.2.3 Reaction of NH₃, NO and O₂. Previous experimental studies showed that the adsorbed NO₂ can provide a stronger electrophilic site and react with NH₃ much more easily than NO,⁵⁸ which results in a higher NO conversion efficiency via the “fast SCR” pathway. However, the detailed reaction path is still unclear, so we further studied its reaction mechanism with the NH₃ molecule based on structure IV′′ in Fig. 6; the specific reaction route and the energy change are shown in Fig. 7. The NH₃ molecule is first dehydrogenated to H, and the dissociated H atom combines with the adsorbed O atom on the surface to form hydroxyl groups and the Fe–OH adduct has a bond distance of 1.812 Å (state II′′′). This step requires an energy of 30 kJ mol⁻¹ to activate the N–H bond. Compared with the reaction path of NH₃ and NO (Scheme 1), it can be observed that the required energy of the NH₃ dissociation in Scheme 3 is much less than that of the NH₃ dissociation and bonding with the surface lattice oxygen in Scheme 1, suggesting that the NH₃ molecule is more prone to direct dissociation and bonding with the surface adsorption of oxygen and further illustrates that the adsorbed oxygen on the α-Fe₂O₃ (001) surface plays an essential role in promoting the SCR deNO_x reaction. Subsequently, free NH₂ and –NO₂ combine to form a –NH₂NO₂ group (state III′′′), which has a stable plane structure similar to –NH₂NO (state IV′, Fig. 5), simultaneously releasing a large amount of energy (ΔE = 288 kJ mol⁻¹), suggesting that the reaction of NH₂ with –NO₂ is much easier than the NH₂NO formation in Scheme 1; this is the reason why the NO_x conversion rate of the SCR reaction in the presence of NO₂ is faster than that of the standard SCR reaction.⁵⁹ This step plays an essential role in the overall reaction by providing energy in the following steps. Fig. 7 shows that the second dehydrogenation of the NH₃ molecule is accompanied by energy dissipation, which indicates that H atoms on –NH₂NO₂ tend to combine with –OH to form –H₂O (state IV′′′). Because the formed –NHNO₂¹ is unstable, we carried out rotation change and structural optimization for it and found that the energy of structure V′′′ is the lowest. There is a small energy barrier (() = 9 kJ mol⁻¹) in the transformation from structure IV′′′ to structure V′′′, indicating that this reaction is relatively likely to occur. In the transition state TS2′′′, a H atom from the –NHNO² approaches an oxygen atom to a distance of 1.738 Å, and at the end of this step (state VI′′′), the N–O bond breaks and the N₂O dissociates out. As for the active energy, this reaction involves the fracture of two bonds, resulting in a higher reaction energy barrier (ΔE = 65 kJ mol⁻¹), and the energy of the final state is lower (91 kJ mol⁻¹) than that of the initial state. The desorption energy of N₂O in structure VI′′′ is relatively small, so its desorption from the system is easy and this mechanism tends to produce N₂O. However, the situation can be different from that of other catalytic systems. For example, N₂O formation over the V₂O₅/AC catalyst during SCR of NO by NH₃ is generally low.⁶⁰ As the reaction progresses, the adsorbed H₂O on the surface of the catalyst dissociates out to form a molecule with energy adsorption (ΔE = 91 kJ mol⁻¹).

Fig. 7 Energy and configuration profiles for the reaction route in Scheme 3. O: red, Mn: gray, H: white, and N: blue.

The N₂O molecule is one of the products of the SCR deNO_x reaction, which can be released into the atmosphere along with the flue gas or be adsorbed on the catalyst surface for further reaction.^61,62 Hence, we calculated the adsorption energy of N₂O adsorbed on the Fe-top site of the α-Fe₂O₃ (001) surface and the energy required for N₂O decomposition to produce N₂. It can be seen from Fig. 7 that the adsorption energy of the N₂O molecule is −21 kJ mol⁻¹, followed by the absorption energy of 66 kJ mol⁻¹ to generate an N₂ molecule (state IX′′′). The activation energy required to break the O–NN bond was 77 kJ mol⁻¹. In the transition state TS3′′′, the distance between N and O was 1.724 Å. By comparison, the adsorption energy of H₂O is much higher than that of N₂O; thus N₂O is unable to compete with H₂O for adsorption at that site. On the basis of CI-NEB results of Scheme 3, we found that N₂ and H₂O molecular dissociation requires a lot of energy, so the rates of both reactions have a great influence on the whole reaction rate, that is to say, the generation of N₂ and H₂O molecules is the RDS of the whole reaction (Scheme 3). The remaining adsorbed O atom can further oxidize NO or combine with the H atoms dissociated from NH₃ to form H₂O, and hydroxyl can also be combined with H atoms to generate H₂O, which can promote the SCR deNO_x reaction. When NH₂ and –NO₂ combine to form the –NH₂NO₂ group, a large amount of energy released can make up for the energy demand of N₂O, N₂ and H₂O generation and dissociation. In this study, we only calculated the reaction route of NH₂ with NO₂(ad) to form H₂O and N₂O. There are other pathways. For instance, NO₂(ad) reacts with OH(ad) to form HNO₂(ad), and then HNO₂(ad) reacts with NH₃(g) to form two H₂O molecules and N₂.⁶³ However, we cannot completely rule out the possibility of other reactions on the surface of α-Fe₂O₃ (001), which requires further study in the future.

4. Conclusions

The adsorption properties and reaction mechanism of NH₃-SCR of NO over α-Fe₂O₃ (001) were explored by DFT calculations. NH₃, NO and O₂ molecules tend to adsorb on the Fe-top site, and the reaction mechanism of NH₃-SCR of NO on the α-Fe₂O₃ (001) surface mainly follows the E–R mechanism. Three elementary reaction schemes were proposed. Scheme 1 is an energetically favourable process, which comprises seven consecutive elementary steps as follows. NH₃ molecule dissociation had an activation barrier of 85 kJ mol⁻¹, and then –NH₂ species interacts with gaseous NO molecule to form FeH–NH₂NO. One H atom of –NH₂NO group dissociates and binds to the O atom on the α-Fe₂O₃ (001) surface to form the second hydroxyl species, and the activation barrier for NH₂NO oxidation is 44 kJ mol⁻¹. Then, –NHNO is rearranged with little energy released. The formation of N₂ releases a large amount of energy. Finally, a H₂O molecule forms and desorbs through recombination of the hydroxyl group and a H atom. The NH₃ dissociation and H₂O generation are the rate-determining step for the reaction of NH₃ with NO (Scheme 1).

NO oxidation (Scheme 2) included two reaction paths. A NO molecule facilely combines with adsorbed O₂. In the following step, Fe–O₂NO may be dissociated to produce a NO₂ molecule (path 1), or may be transformed to Fe–NO₃ (path 2). The Fe–O₂NO group absorbs energy to form a NO₂ molecule, and the NO₂ molecule may adsorb on the α-Fe₂O₃ (001) surface to form a half-bridge nitrite, which has an energy barrier of 30 kJ mol⁻¹. For path 2, the transformation from Fe–O₂NO to Fe–NO₃ has an energy barrier of 122 kJ mol⁻¹. The reaction of NO and O₂ to form a NO₂ molecule is easier than that of Fe–NO₃.

For the reaction of NH₃, NO and O₂ (Scheme 3), the combination of free NH₂ and –NO₂ to form a –NH₂NO₂ group releases a large amount of energy. Dissociation of a H atom and rearrangement of –NHNO₂¹ take place. Then, –NHNO² dissociates to form N₂O, which has an energy barrier of 65 kJ mol⁻¹. The adsorbed H₂O on the surface of the catalyst dissociates out to form a H₂O molecule with energy adsorption. The N₂O molecule is dissociated to form a N₂ molecule and an adsorbed O atom. The generation of N₂ and H₂O molecules is the rate-determining step of the reaction (Scheme 3). In addition, the promoting effect of the superficial adsorbed oxygen in the SCR deNO_x reaction was well illustrated.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We greatly appreciate the financial support provided by the National Natural Science Foundation of China (No. 51676001 and 51376007), the Anhui Provincial Natural Science Foundation (No. 1608085ME104), the Key Projects of Anhui Province University Outstanding Youth Talent (No. gxyqZD2016074 and gxyqZD2017038), and the Funding Projects of Back-up Candidates (No. 2017H131).

References

1. P. H. Abelson, Science, 1985, 230, 617.
2. K. Taylor, Catal. Rev.: Sci. Eng., 1994, 35, 457–481.
3. P. Granger and V. I. Parvulescu, Chem. Rev., 2011, 111, 3155–3207.
4. H. Bosch and F. Janssen, ChemInform, 1988, 19, 369–532.
5. M. A. Larrubia, G. Ramis and G. Busca, Appl. Catal., B, 2001, 30, 101–110.
6. Z. L. Zhao, J. W. Li and B. H. Chen, Xiandai Huagong, 2006, 26, 143–146.
7. J. Han, D. S. Zhang, P. Maitarad, L. Y. Shi, S. X. Cai, H. R. Li, L. Huang and J. P. Zhang, Catal. Sci. Technol., 2015, 5, 438–446.
8. J. Han, J. Meeprasert, P. Maitarad, S. Namuangruk, L. Y. Shi and D. S. Zhang, J. Phys. Chem. C, 2016, 120, 1523–1533.
9. Z. Y. Ren, H. Fan and R. Wang, Catal. Commun., 2017, 100, 71–75.
10. J. Liu, J. Meeprasert, S. Namuangruk, K. W. Zha, H. R. Li, L. Huang, P. Maitarad, L. Y. Shi and D. S. Zhang, J. Phys. Chem. C, 2017, 121, 4970–4979.
11. P. Maitarad, S. Namuangruk, D. S. Zhang, L. Y. Shi, H. R. Li, L. Huang, B. Boekfa and M. Ehara, Environ. Sci. Technol., 2014, 48, 7101–7110.
12. L. J. Yan, Y. Y. Liu, K. W. Zha, H. R. Liu, L. Y. Shi and D. S. Zhang, ACS Appl. Mater. Interfaces, 2017, 9, 2581–2593.
13. M. D. Esrafili, Chem. Phys. Lett., 2018, 695, 131–137.
14. H. B. Liu, Z. X. Zhang, Q. Li, T. H. Chen, C. G. Zhang, D. Chen, C. Z. Zhu and Y. Jiang, Aerosol Air Qual. Res., 2017, 17, 1798–1808.
15. X. L. Mou, B. S. Zhang, Y. Li, L. D. Yao, X. J. Wei, D. S. Su and W. J. Shen, Angew. Chem., Int. Ed., 2012, 51, 2989–2993.
16. S. J. Yang, H. Z. Chang, L. Ma, Z. Qu, N. Q. Yan, C. Z. Wang and J. H. Li, Ind. Eng. Chem. Res., 2013, 52, 5601–5610.
17. K. L. Howard and D. J. Willock, Faraday Discuss., 2011, 152, 135–151.
18. Y. Li, Y. Wan, Y. P. Li, S. H. Zhan, Q. X. Guan and Y. Tian, ACS Appl. Mater. Interfaces, 2016, 8, 5224–5233.
19. S. X. Cai, H. Hu, H. R. Li, L. Y. Shi and D. S. Zhang, Nanoscale, 2016, 8, 3588–3598.
20. Z. F. Yan, S. Shi, Z. Li, Z. J. Zuo, S. Li and X. G. Chen, Catal. Lett., 2017, 147, 1006–1018.
21. W. Y. Song, J. Liu, H. L. Zheng, S. C. Ma, Y. C. Wei, A. J. Duan, G. Y. Jiang, Z. Zhao and E. J. M. Hensen, Catal. Sci. Technol., 2016, 6, 2120–2128.
22. P. Maitarad, J. Meeprasert, L. Y. Shi, J. Limtrakul, D. S. Zhang and S. Namuangruk, Catal. Sci. Technol., 2016, 6, 3878–3885.
23. A. S. Negreira and J. Wilcox, J. Phys. Chem. C, 2016, 117, 24397–24406.
24. L. Arnarson, H. Falsig, S. B. Rasmussen, J. V. Lauritsen and P. G. Moses, J. Catal., 2017, 346, 188–197.
25. R. Wanbayor and V. Ruangpornvisuti, Mater. Chem. Phys., 2010, 124, 720–725.
26. M. Takagi, T. Kawai, M. Soma, T. Onishi and K. Tamaru, J. Catal., 1977, 50, 441–446.
27. M. Inomata, A. Miyamoto and Y. Murakami, J. Catal., 1980, 62, 140–148.
28. F. J. J. G. Janssen, F. M. G. Van den Kerkhof, H. Bosch and J. R. H. Ross, J. Phys. Chem., 1987, 91, 5921–5927.
29. G. Ramis, G. Busca, V. Lorenzelli and P. Forzatti, Appl. Catal., 1990, 64, 243–257.
30. G. T. Went, L. Leu and A. T. Bell, J. Catal., 1992, 134, 479–491.
31. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
32. W. Dong, G. Kresse, J. Furthmüller and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 54, 2157–2166.
33. A. Rohrbach, J. Hafner and G. Kresse, J. Phys.: Condens. Matter, 2003, 15, 979–996.
34. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
35. S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys and A. P. Sutton, Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 57, 1505–1509.
36. O. Bengone, M. Alouani, P. Bloechl and J. Hugel, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 62, 16392–16401.
37. B. Z. Zhu, Q. L. Fang, Y. L. Sun, S. L. Yin, G. B. Li, Z. H. Zi, T. T. Ge, Z. C. Zhu, M. X. Zhang and J. X. Li, J. Mater. Sci., 2018, 53, 11500–11511.
38. A. Rohrbach, J. Hafner and G. Kresse, Phys. Rev. B: Condens. Matter Mater. Phys., 2004, 70, 125426.1–125426.17.
39. L. W. Finger and R. M. Hazen, J. Appl. Phys., 1980, 51, 5362–5367.
40. W. Weiss and W. Ranke, Prog. Surf. Sci., 2002, 70, 1–151.
41. W. Weiss, S. K. Shaikhutdinov, M. Ritter, M. Petersen, F. Wagner, R. Schlögl, M. Scheffler and X. G. Wang, Phys. Rev. Lett., 1998, 81, 1038–1041.
42. R. B. Wang and A. Hellman, J. Chem. Phys., 2018, 148, 094705.
43. L. Bengtsson, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 12301–12304.
44. G. Henkelman, B. Uberuaga and H. Jonsson, J. Chem. Phys., 2000, 113, 9901–9904.
45. C. X. Liu, S. J. Yang, L. Ma, Y. Peng, A. Hamidreza, H. Z. Chang and J. H. Li, Catal. Lett., 2013, 143, 697–704.
46. G. Marbán, T. Valdés-SolíS and A. B. Fuertes, Phys. Chem. Chem. Phys., 2004, 226, 138–155.
47. G. Qi, R. T. Yang and R. Chang, Appl. Catal., B, 2004, 51, 93–106.
48. R. M. Yuan, G. Fu, X. Xu and H. L. Wan, J. Phys. Chem. C, 2014, 115, 21218–21229.
49. A. Vittadini, M. Casarin and A. Selloni, ChemInform, 2005, 36, 1652–1655.
50. Y. Li, Y. Wan, Y. P. Li, S. H. Zhan, Q. X. Guan and Y. Tian, ACS Appl. Mater. Interfaces, 2016, 8, 5224–5233.
51. Z. M. Liu, H. Su, B. H. Chen, J. H. Li and S. I. Woo, Chem. Eng. J., 2016, 299, 255–262.
52. Q. L. Zhang, H. M. Wang, P. Ning, Z. X. Song, X. Liu and Y. K. Duan, Appl. Surf. Sci., 2017, 419, 733–743.
53. D. H. Sun, W. F. Schneider, J. B. Adams and D. Sengupta, J. Phys. Chem. A, 2015, 108, 9365–9374.
54. F. Cao, S. Su, J. Xiang, P. Y. Wang, S. Hu, L. S. Sun and A. C. Zhang, Fuel, 2015, 139, 232–239.
55. F. Cao, J. Xiang, S. Su, P. Y. Wang, L. S. Sun, S. Hu and S. Y. Lei, Chem. Eng. J., 2014, 243, 347–354.
56. J. Xiang, L. L. Wang, F. Cao, K. Qian, S. Su, S. Hu, Y. Wang and L. J. Liu, Chem. Eng. J., 2016, 302, 570–576.
57. L. Arnarson, H. Falsig, S. B. Rasmussen, J. V. Lauritsen and P. G. Moses, J. Catal., 2017, 346, 188–197.
58. M. H. Zhang, X. W. Huang and Y. F. Chen, Phys. Chem. Chem. Phys., 2016, 18, 28854–28863.
59. M. Iwasaki and H. Shinjoh, Appl. Catal., A, 2010, 390, 71–77.
60. P. Li, Q. Y. Liu and Z. Y. Liu, Ind. Eng. Chem. Res., 2011, 50, 1906–1910.
61. Y. Geng, W. P. Shan, S. C. Xiong and F. D. Liu, Catal. Sci. Technol., 2015, 6, 3149–3155.
62. G. Delahay, B. Coq, S. Kieger and B. Neveu, Catal. Today, 1999, 54, 431–438.
63. L. K. Zhao, C. T. Li, Y. Wang, H. Y. Wu, L. Gao, J. Zhang and G. M. Zeng, Catal. Sci. Technol., 2016, 6, 6076–6086.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8cy02080a

---