Locating manganese vanadate phase with PO₄³⁻-modified Mn²⁺–O–V⁵⁺ motifs optimized for catalytic NO_X and poison abatement under oxidative wet conditions†

Seokhyun Lee^a, So Hyeon Park^b and Jongsik Kim*^bc
aAjou Energy Science Research Center, Ajou University, Suwon, 16499, South Korea
^bDepartment of Chemical Engineering (Integrated Engineering Program), Kyung Hee University, Yongin, 17104, South Korea. E-mail: jkim40@khu.ac.kr; Tel: +82-31-201-2444
^cKHU-KIST Department of Converging Science and Technology, Kyung Hee University, Seoul, 02447, South Korea

First published on 24th July 2024

---

---

1 Introduction

An anthropic, wet flue gas often bears NO_X (X = 1–2), O₂, SO₂, and H₃PO₄, where NO_X is infamous as an atmospheric O₃ depleter or acid rain/fine dust generator.^1–4 Of challenge is a high bond-dissociation energy of NO (610 kJ mol⁻¹), which makes it hard to cleave even under a high thermal energy.^5,6 NO, however, can be oxidized to generate paramagnetic NO₂ with thermodynamic spontaneousness, as corroborated by the change in standard Gibbs free energy of −35.3 kJ mol⁻¹ for NO + 1/2O₂ → NO₂.^7–10 Notably, the evolution of NH₄NO₂ is viable on a transition metal oxide surface terminated to afford Brønsted acidic bond (–OH; BA⁻–H⁺) with M_Iⁿ⁺–O–M_Iⁿ⁺ channel being dissected in proximity (M_I/M_II/M_III: metal; n/m/l: formal charge).^11–18 Moreover, the former and latter activates the acidic and redox cycles of selective catalytic NO_X reduction (SCR), respectively, where NH₄NO₂ formation/dissociation are highly feasible (Fig. 1A and B).^11–22 For instance, fragmented M_Iⁿ⁺–O–M_Iⁿ⁺ consists of defective M_I⁽ⁿ⁻¹⁾⁺ vicinal to oxygen vacancy (O_V) and labile oxygen (⁻O–M_Iⁿ⁺; O_LM_Iⁿ⁺ or ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺).^11–18 O_V acts as a reservoir of mobile oxygen (O_M), which can hop on M_I⁽ⁿ⁻¹⁾⁺ defects to regenerate O_LM_Iⁿ⁺ upon O_L consumption vital to recur the redox cycle, whereas O_V is replenished by O₂ of a feed gas (vide infra).^{11–18,23,24}

To expedite the SCR exploiting NH₄NO₂, BA⁻–H⁺ and ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ coordinate with NH₃ and NO, respectively, for yielding BA⁻–NH₄⁺ and NO₂⁻–M_I⁽ⁿ⁻¹⁾⁺ (Fig. 1A).^11–18 These subsequently interact with each other, produce BA⁻–NH₄⁺⋯NO₂⁻–M_I⁽ⁿ⁻¹⁾⁺ (BA⁻–NH₄NO₂–M_I⁽ⁿ⁻¹⁾⁺) via Langmuir–Hinshelwood (LH) model, and release N₂/2H₂O alongside with the evolution of BA⁻ and M_I⁽ⁿ⁻²⁾⁺ (Fig. 1A).^11–18 The former and latter accept H⁺ wandering in the surface and O_M, respectively, to regenerate BA⁻–H⁺ and ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ for the termination of the LH model (Fig. 1A).^11–18 Nonetheless, the LH model is seldom triggered due to the sparsity of M_I to form NO₂⁻–M_I⁽ⁿ⁻¹⁾⁺ except for Mn, to the best of our knowledge.^17,18,25,26 As an alternative to deploy NH₄NO₂ for the SCR activation, BA⁻–H⁺ initially binds with NH₃ to generate BA⁻–NH₄⁺, which in turn interplays with nearby ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ (Fig. 1B).^11–18 This produces BA⁻–NH₄⁺⋯⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ utilized to bind with NO via Eley–Rideal (ER) model for its transformation into N₂/H₂O/BA⁻–H⁺/HO_L–M_I⁽ⁿ⁻¹⁾⁺ (Fig. 1B).^11–18 HO_L–M_I⁽ⁿ⁻¹⁾⁺ experiences dehydration coupled with O_M-assisted oxidation to regenerate ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ with the ER model being terminated (Fig. 1B).^11–18

Of note is that the expedition of the ER model can be more pronounced on the surface with larger quantities of ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ (N_OL) and O_M (N_OM), to which BA⁻–NH₄⁺ and M_I⁽ⁿ⁻²⁾⁺ are in more access for achieving higher collision frequencies among the intermediates involved (Fig. 1B).^11–18 To elevate N_OL/N_OM values, it can be essential to replace one of M_Iⁿ⁺ cations pertaining to M_Iⁿ⁺–O–M_Iⁿ⁺ with M_II^m+. This is because electron (e⁻) affinity for M_Iⁿ⁺/M_II^m+ cations is dissimilar, thus inducing a greater charge separation within M_II^m+–O–M_Iⁿ⁺ than that within M_Iⁿ⁺–O–M_Iⁿ⁺.^27,28 This can cause the cleavage of the former (N_OL/N_OM↑) to be easier than that of the latter (N_OL/N_OM↓).²⁷ In this regard, two channels of M_II^m+–O–M_Iⁿ⁺ and M_III^l+–O–M_Iⁿ⁺ with electro-negativities (EN) of M_II > M_I > M_III should be compared to identify which one is more suitable to accelerate the ER model, whose rate-determining step (RDS) is NO-assisted transition of BA⁻–NH₄⁺⋯⁻O_L–M⁽ⁿ⁻¹⁾⁺ to N₂/H₂O/BA⁻–H⁺/HO_L–M⁽ⁿ⁻¹⁾⁺ (*; Fig. 1B).^11–18 Given the EN values of M_I/M_II/M_III, M_II^m+–O–M_Iⁿ⁺ and M_III^l+–O–M_Iⁿ⁺ can be fragmented to form ⁻O–M_II^m+ (O_LM_II^m+ or ⁻O_L–M_II^(m−1)+) adjacent to O_V/M_I⁽ⁿ⁻¹⁾⁺ defect and ⁻O–M_Iⁿ⁺ (O_LM_Iⁿ⁺ or ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺) vicinal to O_V/M_III^(l−1)+ defect, respectively.^16–18 e⁻ density of ⁻O_L for ⁻O_L–M_II^(m−1)+ is hypothetically lower than that for ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ due to EN values of M_II > M_I. Hence, it was speculated that the amount of e⁻ donated from ⁻O_L to NH₄⁺ for BA⁻–NH₄⁺⋯⁻O_L–M_II^(m−1)+ can be smaller than that for BA⁻–NH₄⁺⋯⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ (*; Fig. 1B).^16–18 This in turn can make the energy required to activate the RDS for the former (E_BARRIER↑) higher than that for the latter (E_BARRIER↓).^16–18 Moreover, e⁻ density of M_I⁽ⁿ⁻¹⁾⁺ defect was conjectured to be higher than that of M_III^(l−1)+ defect (EN values of M_I > M_III). This thus can render higher intimacy of O_M with M_I⁽ⁿ⁻¹⁾⁺ defect than that with M_III^(l−1)+ counterpart, potentially resulting in a lower O_M mobility for the fragmented M_II^m+–O–M_Iⁿ⁺ (O_M mobility↓) compared to that for the fragmented M_III^l+–O–M_Iⁿ⁺ (O_M mobility↑) in the redox cycle of the ER model (Fig. 1B).^16–18 Besides, e⁻ intimacy of M_I⁽ⁿ⁻¹⁾⁺ > M_III^(l−1)+ tentatively induces a higher interaction between O²⁻ of H₂O being fed and defective M_I⁽ⁿ⁻¹⁾⁺ (or O_V) for the former (H₂O resistance↓) than that for the latter (H₂O resistance↑).^16–18 On the whole, M_III^l+–O–M_Iⁿ⁺ can outperform M_II^m+–O–M_Iⁿ⁺ with regard to E_BARRIER, O_M mobility, and H₂O resistance. Therefore, considering the EN values of 3d-block transition metals in frequent use as the components of the SCR catalysts (Mn (3.46 eV) < V (3.64 eV) < Fe/Co/Ni/Cu (4.03–4.48 eV) in Mulliken scale), Mn and V were rationally chosen as M_III and M_I for M_III^l+–O–M_Iⁿ⁺, respectively, herein.²⁹

Of additional note is that the selection of +2/+5 as the formal charges of Mn(l+)/V(n+) results in the creation of Mn²⁺–O–V⁵⁺ motif used to construct manganese vanadates (Mn_XV₂O_X+5) such as equimolar Mn:V of monoclinic for X = 1 and Mn-rich Mn:V of monoclinic for X = 2/orthorhombic for X = 3 (Fig. 1D–F).^30–32 Interestingly, the transition metal vanadates explored by us to-date (M_XV₂O_X+5; X = 1–3 for Ni and Cu; X = 1 for Mn and Co) possess V-rich surfaces.^11,14 Mn²⁺–O–Mn²⁺ motifs may thus be sparse across the Mn_XV₂O_X+5 surfaces and were excluded from the discussion. Meanwhile, V⁵⁺–O–V⁵⁺ motifs may be plentiful throughout the Mn_XV₂O_X+5 surfaces, yet, are featured by quite a smaller charge separation than Mn²⁺–O–V⁵⁺ counterparts, thus permitting us to rule out V⁵⁺–O–V⁵⁺ from the discussion.^27,28 Again, Mn²⁺–O–V⁵⁺ channel can be fragmented to evolve Mn⁺, O_V, and ⁻O–V⁵⁺ (O_LV⁵⁺ or ⁻O_L–V⁴⁺).^16–18 It should be stressed that Mn⁺ is defective/Lewis acidic (LA) to bear SO_A²⁻ (A = 3–4) and PO₄³⁻ functionalities via modification with SO₂ and H₃PO₄ poisons, respectively, in an O₂-ample feed gas.^11–18,33 The surface SO_A²⁻ and PO₄³⁻ modifiers afford terminal S⁴⁺–O²⁻/S⁶⁺–O²⁻ and P⁵⁺–O²⁻ (BA⁻), respectively, all of which are readily protonated to function as BA⁻–H⁺ bonds (Fig. 1C).^{11–18,33–35} This indicates that the SO_A²⁻/PO₄³⁻-modified Mn_XV₂O_X+5 surfaces can be LA-lean and BA⁻–H⁺-rich, thus leading us to primarily consider the ER model mediated by BA⁻–H⁺ (Fig. 1B).^11–18,33 In addition, we demonstrated the amount (N_BA⁻–H⁺) and strength (E_BA⁻–H⁺) of BA⁻–H⁺ bonds for a protonated SO_A²⁻ functionality can be dictated by its A value, binding array (mono- or bi-dentate), and the kind of a Mn_XV₂O_X+5 (transition metal vanadate) host on which a SO_A²⁻ guest is grafted.¹⁴ Our validation stated above is hypothetically adaptable to a protonated PO₄³⁻ guest (Fig. 1C), whose significance and/or roles in activating the BA⁻–H⁺-mediated ER model on Mn_XV₂O_X+5 hosts have never been studied so far.

Here we thus postulated that the traits of protonated PO₄³⁻ (BA⁻–H⁺; N_BA⁻–H⁺/E_BA⁻–H⁺) and neighboring O_M (∼O_V)/O_L (N_OM/OL/E_OM/OL) can vary with the change in the X values for PO₄³⁻-modified Mn_XV₂O_X+5 architectures (Fig. 1D–F and Table S3†), whose secondary building units impart distinct Mn²⁺–O–V⁵⁺ connectivities with e⁻ migrating from Mn²⁺ to V⁵⁺ (EN values of Mn < V). Intact Mn_XV₂O_X+5 architectures have octahedral [Mn²⁺–(O²⁻)₆]¹⁰⁻ secondary building units, in which Mn²⁺ centers are coordinatively-saturated, whereas the numbers of Mn²⁺–O–V⁵⁺ channels in a per-Mn²⁺ basis (N_{V⁵⁺–O–Mn²⁺}) are eight for X = 1–2 and six for X = 3 (Fig. 1D–F, Tables S1 and S3†).^30–32 Hence, in a Mn²⁺–O–V⁵⁺ channel, Mn²⁺ can donate a greater amount of e⁻ to V⁵⁺ at a smaller N_{V⁵⁺–O–Mn²⁺}, tentatively enabling Mn²⁺ to be more e⁻-deficient at a larger X. This hypothetically shortens the length of Mn²⁺–O and elevates its binding strength at a smaller N_{V⁵⁺–O–Mn²⁺}. This suggests that the fragmentation of a Mn²⁺–O–V⁵⁺ channel into Mn⁺/O_V/⁻O–V⁵⁺ (⁻O_L–V⁴⁺) may be more reluctant at a smaller N_{V⁵⁺–O–Mn²⁺} and thus potentially reduces N_OV/N_OM/N_OL at a larger X (N_OV/N_OM/N_OL↓ at X of 1/2 → 3). N_OV reduction may also incur the decrease in the spots used to coordinate with protonated PO₄³⁻ functionalities with their P⁵⁺–O²⁻–H⁺ bonds accessible to NH₃ (N_NH3 ∼ N_BA⁻–H⁺↓ at X of 1/2 → 3), resulting in the decrease in the number of BA⁻–NH₄⁺ available to form BA⁻–NH₄⁺⋯⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ prior to its collision with NO . This also suggests that if Mn²⁺–O–V⁵⁺ dissection is indispensable, the binding strength of O_M with Mn⁺ (E_OM) may be higher at a smaller N_{V⁵⁺–O–Mn²⁺}, thus potentially lowering O_M mobility at a larger X (E_OM↑/O_M mobility↓ at X of 1/2 → 3).^16–18 This further suggests that the binding intimacy of Mn⁺ with O²⁻ of P⁵⁺–O²⁻ for a PO₄³⁻ functionality may be greater at a smaller N_{V⁵⁺–O–Mn²⁺}, where O²⁻ of surface-unbound P⁵⁺–O²⁻ for a PO₄³⁻ modifier may be more e⁻-deficient and less proton (H⁺)-affinitive at a larger X. Consequently, H⁺ bound to P⁵⁺–O²⁻ (BA⁻–H⁺) for a protonated PO₄³⁻ modifier may coordinate with NH₃ more readily at a smaller N_{V⁵⁺–O–Mn²⁺}, tentatively resulting in the elevation of binding intimacy between BA⁻–H⁺ and NH₃ at a larger X (E_NH3 ∼ E_BA⁻–H⁺↑ at X of 1/2 → 3).^16–18 This may also hinder NO-assisted BA⁻–NH₄⁺⋯⁻O_L–V⁴⁺ dissociation (RDS) with the increase in E_BARRIER at a larger X (E_BARRIER↑ at X of 1/2 → 3).^16–18

On the other hand, defect-free Mn_XV₂O_X+5 architectures also possess octahedral [V⁵⁺–(O²⁻)₆]⁷⁻ sub-unit for X = 1 and tetrahedral [V⁵⁺–(O²⁻)₄]³⁻ sub-units for X = 2–3, among which V⁵⁺ centers are open to H₃PO₄/O₂ at X ≥ 2 (Fig. 1D–F, Tables S2 and S3†).^30–32 This can tentatively make it plausible that the number of P⁵⁺–O²⁻–H⁺ bonds for protonated PO₄³⁻ functionalities may be larger in X = 2–3 than in X = 1, if their binding configurations are identical. This is again due potentially to a larger amount of spots available to immobilize protonated PO₄³⁻ modifiers for X ≥ 2 than those for X = 1 . Moreover, the quantities of Mn²⁺–O–V⁵⁺ channels in a per-V⁵⁺ basis (N_{Mn²⁺–O–V⁵⁺}) are four for X = 1, six for X = 2, and nine for X = 3 (Fig. 1D–F, Tables S2 and S3†).^30–32 Therefore, in a Mn²⁺–O–V⁵⁺ channel, V⁵⁺ can accept a greater quantity of e⁻ from Mn²⁺ at a larger N_{Mn²⁺–O–V⁵⁺}, potentially enabling V⁵⁺ to be more e⁻-abundant at a greater X. This tentatively elongates the distance of ⁻O–V⁵⁺ (⁻O_L–V⁴⁺), weakens its binding affinity at a larger N_{Mn²⁺–O–V⁵⁺}, yet, can make ⁻O_L (⁻O) of ⁻O_L–V⁴⁺ (⁻O–V⁵⁺) more nucleophilic at a larger X. ⁻O_L of ⁻O_L–V⁴⁺ for a fragmented Mn²⁺–O–V⁵⁺ may thus bind with H⁺ of BA⁻–NH₄⁺ more rigidly at a larger N_{Mn²⁺–O–V⁵⁺} (E_OL↑ at X of 1 → 3) and hypothetically expedite the disintegration of BA⁻–NH₄⁺⋯⁻O_L–V⁴⁺ upon its interaction with NO (RDS) by lowering E_BARRIER at a larger X (E_BARRIER↓ at X of 1 → 3).^16–18 V⁴⁺ of ⁻O_L–V⁴⁺ for a fragmented Mn²⁺–O–V⁵⁺ can also affect the properties of O_M, whose mobility may be promoted at a larger N_{Mn²⁺–O–V⁵⁺} due possible to a greater repulsion between O_M and e⁻-abundant V⁴⁺ at a larger X (E_OM↓/O_M mobility↑ at X of 1 → 3).^16–18 Furthermore, the binding strength of V⁴⁺ with O²⁻ of P⁵⁺–O²⁻ for a PO₄³⁻ functionality may be lower at a larger N_{Mn²⁺–O–V⁵⁺}, potentially resulting from a greater e⁻ abundance of V⁴⁺ at a larger X. This in turn tentatively causes O²⁻ of surface-unbound P⁵⁺–O²⁻ for a PO₄³⁻ modifier (BA⁻) to be e⁻-rich and H⁺-affinitive at a larger N_{Mn²⁺–O–V⁵⁺}. This may reduce the binding affinity of BA⁻–H⁺ for NH₃ at a larger X (E_NH3 ∼ E_BA⁻–H⁺↓ at X of 1 → 3), potentially helping reduce E_BARRIER at a larger X (E_BARRIER↓ at X of 1 → 3).^16–18

The inspection on local Mn²⁺–O–V⁵⁺ connectivities for Mn_XV₂O_X+5 architectures provides the predicted hierarchies on N_BA⁻–H⁺/E_BA⁻–H⁺/N_OL/E_OL/N_OM/E_OM/O_M mobility/E_BARRIER/ values versus X, most of which are contradictory from each other except for those on N_OL/E_OL/N_OM versus X (Table S3†). These will remain the X value most proper to activate the wet SCR difficult-to-determine, unless being elaborated via control experiments. Moreover, PO₄³⁻-modified Mn_XV₂O_X+5 surfaces exposed to a wet, SO₂-containing feed gas can be poisoned by ammonium (bi-)sulfate (AS/ABS), yet, can activate AS/ABS pyrolysis, whose and E_BARRIER are dominated by N_BA⁻–H⁺ and E_BA⁻–H⁺/hydrophilicity of the surface, respectively , as detailed below (Fig. 8B).^14,16–18 The predicted trends on N_BA⁻–H⁺/E_BA⁻–H⁺ values versus X in the light of N_{V⁵⁺–O–Mn²⁺} values are opposite to those in the aspect of N_{Mn²⁺–O–V⁵⁺} counterparts (Table S3†), thus leaving it challenging to confirm the X value optimized for accelerating AS/ABS pyrolysis. All of these propelled us to experimentally compare the generic, acidic (N_BA⁻–H⁺/E_BA⁻–H⁺), redox (N_OM/E_OM/O_M mobility/N_OL/E_OL), and hydrophobic nature of PO₄³⁻-modified Mn_XV₂O_X+5 surfaces. Their catalytic properties in the SCR or AS/ABS pyrolysis were also inspected in reaction-limited and diffusion-limited domains for comparison in terms of activities/kinetic parameters and performance, respectively. The optimum Mn_XV₂O_X+5 surface subjected to PO₄³⁻ modification was further improved using promotive Sb₂O₅ for comparison with Sb₂O₅-promoted Mn₁V₂O₆ on TiO₂ subjected to SO_A²⁻ modification (Mn₁–Sb–S) or commercial control of WO₃-promoted V₂O₅ on TiO₂ (V₂O₅–WO₃).

2 Results and discussion

2.1 General properties of the catalysts

Mn_XV₂O_X+5 building units were dispersed on NH₃/SO₂-resistant, mesoporous TiO₂ (anatase) support via wet impregnation–calcination techniques, whose procedures were identical across the resulting Mn_X catalysts except for varying the molar ratios of Mn/V precursors used to regulate the X values.^11,14–18 The synthetic protocols stated above served to rigorously test the feasibilities of the trends anticipated for the surface/catalytic properties of the Mn_X catalysts versus X (vide supra), most of which were out-of-consensus in the lights of N_{V⁵⁺–O–Mn²⁺}/N_{Mn²⁺–O–V⁵⁺} values for intact Mn_XV₂O_X+5 architectures. Notably, the Mn_X catalysts were synthesized with an aim to bear bulk V contents of 2.0 wt% suitable to activate the SCR or AS/ABS pyrolysis at ≤400 °C, as reported elsewhere.^{11,12,14–18,33}

Meanwhile, (NH₄)₂HPO₄ bears P⁵⁺–O²⁻ bonds prone to bind with defective metal cations adjacent to O_V species (Mn⁺/V⁵⁺ of a fragmented Mn²⁺–O–V⁵⁺ for X = 1–3) or intrinsically open metals (V⁵⁺ of [V⁵⁺–(O²⁻)₄]³⁻ for X ≥ 2) on/near the surface via mono-/bi-dentate binding modes.^12,33,36 Moreover, (NH₄)₂HPO₄ on the surface can undergo deamination and/or dehydration under oxidative environments at ≤500 °C, leading to its transformation into surface PO₄³⁻ functionality acting as BA⁻ (Fig. 1C).^36–38 This is in contrast to surface-unbound (NH₄)₂HPO₄, whose oxidative transition is from H₃PO₄ via deamination at <200 °C, 1/2(H₄P₂O₇) via dehydration at 200–300 °C, to gaseous P₄O₁₀ via poly-condensation/sublimation at ≤500 °C.^39,40 In this regard, Mn_X catalysts were impregnated with (NH₄)₂HPO₄, whose quantity was determined based on bulk molar ratios of S to metal (0.2–0.3) for SO_A²⁻-modified transition metal vanadates to provide a host of merits in accelerating the SCR, as we verified previously.^14–18 The resulting (NH₄)₂HPO₄-impregnated Mn_X catalysts were then calcined identically to those prior to (NH₄)₂HPO₄ impregnation. This led to the formation of Mn_X–P catalysts, whose (protonated) PO₄³⁻ contents (N_BA⁻–H⁺) could be distinct and rely heavily on the amounts of defective Mn⁺ or defective/open V⁵⁺ species for Mn_XV₂O_X+5 architectures.

Furthermore, TiO₂-supported Sb₂O₅ (Sb₂O₅/TiO₂) was synthesized in such a way to bear bulk Sb content of 3.0 wt%, leading frequently to the promotion of the SCR activities, AS/ABS pyrolysis activities, or H₂O/SO₂ resistance for bunched metal vanadates we have discovered to-date.^{11,12,14,15,17,18,41–43} Notably, Mn₁–P outperformed Mn₂–P/Mn₃–P in promoting acidic and/or redox properties pivotal to catalyze the SCR, which was demonstrated below in substantial detail. Mn₁–Sb–P was thus synthesized by positioning Mn₁V₂O₆ architectures adjacent to Sb₂O₅ promoters of Sb₂O₅/TiO₂ and subjected to PO₄³⁻ functionalization under the same conditions to those used to synthesize Mn₁–P except for the substitution of Sb₂O₅/TiO₂ for TiO₂.

The macroscopic morphologies of the catalysts were observed using scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX) mapping images, where the catalysts were comprised of micron-sized, irregular-shaped chunks with all the components (Mn, V, Sb, and P) active to the SCR or AS/ABS pyrolysis being highly scattered on the surfaces (Fig. S1†). The microscopic morphologies of the catalysts were also explored using high-resolution transmission electron microscopy (HRTEM) images (Fig. 2A–D), in which 30–50 nm-sized TiO₂ particulates were aggregated. This was evidenced by the lattice fringes with d spacings of 2.37 Å and 3.52 Å, which were indexed to surface (004) and (101) facets for tetragonal TiO₂, respectively. The textural properties of the catalysts were then inspected using their type IV N₂ isotherms at −196 °C (not shown), from which the catalysts were identified to possess meso-porosities accessible to N₂, as gauged by Brunauer–Emmett–Teller surface areas (S_BET,N2) of ∼76.5 m_N2² g_CAT⁻¹ and Barrett–Joyner–Halenda pore volumes (V_BJH,N2) of ∼0.3 cm_N2³ g_CAT⁻¹ (Table S4†).

The bulk and surface compositional features of the catalysts were further investigated using inductively coupled plasma (ICP) combined with optical emission spectroscopy (OES) or atomic absorption spectroscopy (AAS), with which the catalysts were found to possess bulk V contents of 2.0 wt%, as-targeted. The catalysts were also verified to have such bulk molar ratios of Mn to V (Mn/V) that were in close accordance to those utilized to define Mn_XV₂O_X+5 building units (e.g., Mn/V of 0.5 for Mn₁ and Mn₁–Sb–P; 1.4 for Mn₃–P; Table S4†), which might suggest the successful Mn_XV₂O_X+5 deposition on TiO₂ surfaces.^30–32

Interestingly, the magnitude of the bulk Mn/V values for the catalysts was around twice that of their corresponding surface Mn/V values, as assessed via X-ray photoelectron (XP) spectroscopy or EDX mapping (Table S4†). This could originate from V-rich traits of the catalyst surfaces, as also found in other transition metal vanadates reported so far.^11,14 This might justify the exclusion of Mn²⁺–O–Mn²⁺ channels inherent to PO₄³⁻-modified Mn_XV₂O_X+5 architectures from the predictions with regard to their acidic/redox/hydrophobic features. Conversely, the molar ratios of P relative to metals (P/metal) for the catalysts in the surface scale (0.7) were 2-4-fold higher than those for the corresponding catalysts in the bulk scale (0.2–0.3), as shown in Table S4.† This indicated that PO₄³⁻ functionalities mainly served to modify the surface spots such as defective Mn⁺ or defective/open V⁵⁺ species. Apparently, surface molar ratios of metal to V (metal/V) increased in the sequence of Mn₁–P (0.2) < Mn₂–P (0.4) < Mn₁–Sb–P (0.6) ∼ Mn₃–P (0.7) with their surface P/metal values being alike (Table S4†). These suggested that the quantity of (protonated) PO₄³⁻ modifiers (N_BA⁻–H⁺) could be larger in the order of Mn₁–P < Mn₂–P < Mn₁–Sb–P ∼ Mn₃–P. Again, N_BA⁻–H⁺ values of the catalysts were speculated to be inversely-proportional and proportional to X values in the light of defective Mn⁺ and open V⁵⁺ species, respectively. Nonetheless, open V⁵⁺ species were identified to outweigh defective Mn⁺ counterparts in dictating N_BA⁻–H⁺ values of the Mn_X–P catalysts with defective Sb cations on terminated Sb₂O₅ surface acting as additional grafting points of (protonated) PO₄³⁻ functionalities for Mn₁–Sb–P.

The bulk crystallographic characteristics of the catalysts were investigated using their X-ray diffraction (XRD) patterns (Fig. S2†), across which the bulk diffractions were assigned to those for tetragonal TiO₂ (anatase) only and did not match those for its polymorph (rutile), whereas the bulk facets corresponding to those for Mn_XV₂O_X+5 or Sb₂O₅ were absent due in part to the inclusion of small Mn/V/Sb contents (≤3.0 wt%).^{11,14–18,42} The surface crystallographic traits of the catalysts were thereby explored using their selected area electron diffraction (SAED) patterns (Fig. 2E–H), where the surface diffractions indexed to those for tetragonal TiO₂ were consistently observed. The other surface facets found in the SAED patterns were assigned to those of monoclinic Mn₁V₂O₆ for Mn₁–P, monoclinic Mn₂V₂O₇ for Mn₂–P, orthorhombic Mn₃V₂O₈ for Mn₃–P, and monoclinic Mn₁V₂O₆ coupled with cubic Sb₂O₅ for Mn₁–Sb–P, while collateral surface diffractions of Mn₃(PO₄)₂ or V₃(PO₄)₅ were un-indexed (Fig. 2E–H). All the characterization results could allow for the conclusion that the catalysts were synthesized as-aimed and adequate to strictly rank their surface/catalytic properties in terms of X values.

2.2 Acidic properties of the catalysts

The catalyst surfaces should bear Mn²⁺ and V⁵⁺ only, if their Mn_XV₂O_X+5 architectures were intact (Fig. 1D–F).^14–18 The surfaces, however, were verified to have multiple facets upon termination (via SAED; Fig. 2E–H), thus tentatively enabling to afford Mn^3+/4+ and V^3+/4+, all of which could be regarded as LA defects accessible to O²⁻ of P⁵⁺–O²⁻ for PO₄³⁻ (BA⁻) and/or O_M species.^{11,12,14–18,41–43} In this regard, metal phases present in the catalyst surfaces and their compositions were examined using XP spectroscopy.

The XP spectra of the catalysts in the Mn 2p regimes were de-convoluted into six sub-bands upon the exclusion of satellites with binding energies centered at 645.0–648.0 eV (Fig. S3 and Table S5†).^13,14,44,45 These sub-bands belonged to the Mn 2p_1/2 or Mn 2p_3/2 regions, either of which afforded three sub-bands assigned to Mn²⁺, Mn³⁺, and Mn⁴⁺ with binding energies centered at 652.0–652.7 eV (Mn 2p_1/2)/640.6–641.3 eV (Mn 2p_3/2), 653.4–654.1 eV (Mn 2p_1/2)/641.7–642.4 eV (Mn 2p_3/2), and 655.2–655.9 eV (Mn 2p_1/2)/643.2–643.9 eV (Mn 2p_3/2), respectively.^13,14,44,45 Defective Mn^3+/4+ concentrations increased in the order of Mn₁–P (58.8%) < Mn₁–Sb–P (63.9%) < Mn₂–P ∼ Mn₃–P (71.6%), which was against our prediction of a smaller N_OV at a larger X in the aspect of N_{V⁵⁺–O–Mn²⁺} values. This could partially originate from fragmented Mn²⁺–O–Mn²⁺ motifs, whose significance was likely to become greater at X ≥ 2 than at X = 1, yet, was in partial line with our anticipation concerning the numbers of PO₄³⁻ anchoring points using metal/V and P/metal values in addition to defective Sb cations for the catalyst surfaces (N_BA⁻–H⁺ of Mn₁–P < Mn₂–P < Mn₁–Sb–P ∼ Mn₃–P). In addition, binding energies of defective Mn^3+/4+ phases were elevated in the order of Mn₁–Sb–P < Mn₁–P ∼ Mn₂–P < Mn₃–P (e.g., Mn³⁺ of 641.7 eV for Mn₁–Sb–P; 642.0 eV for Mn₁–P/Mn₂–P; 642.4 eV for Mn₃–P in the Mn 2p_3/2 domains). This indicated that defective Mn cations were more e⁻-deficient at X = 3 than X = 1–2, thereby potentially resulting in E_OM↑/O_M mobility↓/E_BA⁻–H⁺↑ at X of 1/2 → 3, as also predicted in the light of N_{V⁵⁺–O–Mn²⁺} values.^16–18 Interestingly, albeit Sb possesses the highest EN (4.85 eV in Mulliken scale) among all the active components, Sb⁵⁺ of Sb₂O₅ was likely to donate e⁻ to defective Mn^3+/4+ of proximal Mn₁V₂O₆ and might cause O_M mobilities of Mn₁–Sb–P > Mn₁–P and E_BA⁻–H⁺ values of Mn₁–Sb–P < Mn₁–P.²⁹

The XP spectra of the catalysts in the V 2p_3/2 regimes were also curve-fitted to three sub-bands, which were assigned to V³⁺, V⁴⁺, and V⁵⁺ with binding energies centered at 515.0–515.9 eV, 516.0–516.9 eV, and 516.7–517.6 eV, respectively (Fig. S4 and Table S5†).^14,15,46,47 Noteworthily, Mn₂V₂O₇/Mn₃V₂O₈ bear intrinsically open V⁵⁺ species, which were combined with defective V^3+/4+ to make Lewis acidic V phases challenging to quantify using the XP spectra of Mn₂–P/Mn₃–P.^30–32 Nevertheless, binding energies of V^3+/4+/5+ phases increased in the sequence of Mn₃–P < Mn₂–P < Mn₁–P < Mn₁–Sb–P (e.g., V³⁺ of 515.0 eV for Mn₃–P; 515.3 eV for Mn₂–P; 515.6 eV for Mn₁–P; 515.9 eV for Mn₁–Sb–P). This indicated that Lewis acidic V cations were more e⁻-abundant at a larger X, thereby potentially causing E_OM↓/O_M mobility↑/E_BA⁻–H⁺↓ at X of 1 → 3, as also expected in the aspect of N_{Mn²⁺–O–V⁵⁺} values.^16–18 In contrast to hypothetic e⁻ donation from Sb⁵⁺ of Sb₂O₅ to defective Mn^3+/4+ of vicinal Mn₁V₂O₆, Sb⁵⁺ of Sb₂O₅ might extract e⁻ from Lewis acidic V cations of neighboring Mn₁V₂O₆ and thereby was likely to bring about O_M mobilities of Mn₁–Sb–P < Mn₁–P and E_BA⁻–H⁺ values of Mn₁–Sb–P > Mn₁–P. Apparently, the trends on E_OM/O_M mobility/E_BA⁻–H⁺ values of the catalysts versus X predicted using their XP spectra in the Mn 2p regimes were opposite to those in the V 2p_3/2 domains.

Of note was that the binding arrays of surface PO₄³⁻ modifiers were arduous to clarify experimentally owing to the lack of analytic techniques capable of doing so, as far as we know.^12,48,49 Nonetheless, PO₄³⁻ modifiers were reported to bind with metal oxide surfaces via mono-/bi-dentate arrays with energetic spontaneousness, as proved via density functional theory calculations.^12,50–52 Moreover, the catalyst surfaces synthesized herein were highly defective, V-rich, and featured by Mn²⁺–O–Mn²⁺/Mn²⁺–O–V⁵⁺/V⁵⁺–O–V⁵⁺ motifs with diverse coordination environments.^30–32 All of these were gathered to make it sound that the catalyst surfaces bore PO₄³⁻ functionalities with multiple (mono-/bi-)binding modes.

Of additional note was that PO₄³⁻ (BA⁻) functionality with a higher tendency to protonation can provide the surface with a larger N_BA⁻–H⁺. The catalysts were thus examined with regard to PO₄³⁻ modifier and its protonated analogues using their XP spectra in the P 2p_3/2 regimes (Fig. 3A–D and Table S5†). These were de-convoluted into three sub-bands, whose assignment were PO₄³⁻, HPO₄²⁻, and H₂PO₄⁻ phases with binding energies centered at 131.7 eV, 132.7 eV, and 133.6 eV, respectively.^12,33,53 Interestingly, the concentrations of PO₄³⁻/HPO₄²⁻/H₂PO₄⁻ phases were comparable across the surfaces, which could suggest that their PO₄³⁻ functionalities tended to be protonated in similar fashion. Moreover, the concentrations of HPO₄²⁻ (8.1–9.6%) and H₂PO₄⁻ (79.0–80.7%) were far higher than that of PO₄³⁻ across the surfaces. This could indicate that the surfaces bore PO₄³⁻ functionalities highly susceptible to protonation. The catalysts were then examined using ³¹P magic angle spinning nuclear magnetic resonance (³¹P MAS NMR) spectroscopy to further inspect the type of (protonated) PO₄³⁻ functionalities and their compositions. ³¹P MAS NMR spectra of the catalysts were curve-fitted to three sub-bands with chemical shifts centered at −24.6 ppm, −17.6 ppm, and −10.6 ppm (Fig. 3E–H and Table S6†).^12,33,54 These could be indexed to PO₄³⁻, HPO₄²⁻, and H₂PO₄⁻, respectively, whose concentrations were almost identical across the catalysts with the main possession of HPO₄²⁻ (7.7–9.8%) and H₂PO₄⁻ (78.4–80.8%).^12,33,54 This could also demonstrate the similar inclination of the catalyst surfaces with regard to the protonation of their PO₄³⁻ functionalities. In sum, the surfaces were likely to bind with PO₄³⁻ functionalities via mono-/bi-dentate configurations, across which their liability to protonation was barely distinct. These could leave the kind of Mn_XV₂O_X+5 hosts as a prime dictator of Brønsted acidities (N_BA⁻–H⁺/E_BA⁻–H⁺) for (protonated) PO₄³⁻ guests.

CO-pulsed chemisorption served to quantify the amounts of LA species (e.g., defective Mn⁺ and/or defective/open V⁵⁺) accessible to CO in a per-gram of the catalysts at 50 °C (N_CO; Table S4†).^11–18 N_CO values of the catalysts prior to PO₄³⁻ modification increased in the sequence of Mn₁ (2.1 × 10⁻¹ μmol_CO g_CAT⁻¹) < Mn₁–Sb (2.8 × 10⁻¹ μmol_CO g_CAT⁻¹) < Mn₂ ∼ Mn₃ (5.2 × 10⁻¹ μmol_CO g_CAT⁻¹). This again could stem from the inclusion of defective Sb cations for Mn₁–Sb and innately open V⁵⁺ centers for Mn₂/Mn₃. The N_CO values of the catalysts were reduced by 0.4 × 10⁻¹–1.2 × 10⁻¹ μmol_CO g_CAT⁻¹ post PO₄³⁻ functionalization, during which a portion of defective/open LA species served to coordinate with PO₄³⁻ modifiers, while leaving quite a few defective metal cations adjacent to O_V species.^12,33,36

In addition, NH₃-temperature-programmed desorption (NH₃-TPD) experiments were performed by initially chemisorbing NH₃ molecules on the surfaces at 50 °C prior to their liberation from the surfaces at ≤700 °C with the ramping rate (β) of 10 °C min⁻¹ (Fig. S5†).^11–18,33 The objective of NH₃-TPD runs was to assess the numbers of LA species and BA⁻–H⁺ bonds accessible to NH₃ in a per-gram of the catalysts (N_NH3) at 50 °C with the use of the areas under the resulting NH₃-TPD profiles (thermal conductivity detector (TCD) signal versus temperature).^11–18,33 It should be stressed that the majority of NH₃-accessible sites for the PO₄³⁻-modified catalysts could correspond to their BA⁻–H⁺ bonds. This was because of LA consumption via PO₄³⁻ occupation along with the impartation of additional BA⁻–H⁺ species via protonation of terminal P⁵⁺–O²⁻ bonds for PO₄³⁻ modifiers, as proved in a spectrum of reports on metal vanadates/metal oxide composites subjected to SO_A²⁻/PO₄³⁻ modification.^11–18,33 N_NH3 (∼N_BA⁻–H⁺) values of the PO₄³⁻-modified catalysts increased in the sequence of Mn₁–P ∼ Mn₂–P (290–300 μmol_NH3 g_CAT⁻¹) < Mn₁–Sb–P ∼ Mn₃–P (330–340 μmol_NH3 g_CAT⁻¹), whose hierarchy was in partial line with that using their surface P/metal and metal/V values. This might prove the centrality of X value-directed properties for defective/open metal cations in regulating the nature of (protonated) PO₄³⁻ functionalities.

NH₃-TPD technique additionally served to unveil the significance of X values (type of Mn_XV₂O_X+5 architectures) in determining N_BA⁻–H⁺/E_BA⁻–H⁺ values for the catalysts in use for activating the acidic cycle of the low-temperature SCR (≤250 °C). NH₃-TPD runs on the PO₄³⁻-modified catalysts were conducted identically to those discussed above except for changing the temperature utilized to chemisorb NH₃ molecules on the surfaces (T_NH3) from 50 °C to 220 °C with the β variation of 10–30 °C min⁻¹ (Fig. S6, S7 and Table S7†).^11–18,33 The areas under the resulting NH₃-TPD profiles of the catalysts were used to evaluate their N_NH3 values at 220 °C, whose magnitude was around a third of that at 50 °C, originating from a higher T_NH3 (50 → 220 °C) to exert higher thermal energies on the surfaces (Fig. 4A and S5†).^12,15,18 Conversely, the hierarchy on the N_NH3 values of the catalysts was irrespective of the change in T_NH3 values (Mn₁–P ∼ Mn₂–P (90–100 μmol_NH3 g_CAT⁻¹) < Mn₁–Sb–P ∼ Mn₃–P (120–130 μmol_NH3 g_CAT⁻¹)). This suggested the selection of a greater X value or the deposition of defective Sb cations on the surface could be desired to achieve a greater N_NH3 value.

NH₃-TPD profiles were then curve-fitted to three sub-bands, each of which provided a set of temperatures with maximum NH₃ desorption (T_MAX) when β values varied from 10 °C min⁻¹ to 30 °C min⁻¹.^11–18,33 ln (β/T_MAX²) values were then related with their corresponding 1/T_MAX values for a sub-band (Fig. 4A and Table S7†), whose slope corresponds to −E_NH3/R according to the TPD theory (R: the ideal gas constant).^11–18,33 E_NH3 (∼E_BA⁻–H⁺) values were elevated in the sequence of Mn₁–Sb–P (22.5 kJ mol_NH3⁻¹) < Mn₁–P ∼ Mn₂–P (25.3–27.8 kJ mol_NH3⁻¹) < Mn₃–P (30.5 kJ mol_NH3⁻¹), whose rank was in accordance with that anticipated in the aspect of N_{V⁵⁺–O–Mn²⁺} values or from the XP spectral results in the Mn 2p domains.^16–18 This suggested that albeit with the major possession of open/defective V cations on the surfaces (via surface Mn/V values), defective Mn cations should not be disregarded in determining E_BA⁻–H⁺ values of the catalysts. Importantly, the examination on the acidic traits of the catalysts demonstrated their marked reliance on the type of Mn_XV₂O_X+5 hosts, throughout which defective Mn and defective/open V cations were highly coupled to afford (protonated) PO₄³⁻ guests with dissimilar N_BA⁻–H⁺/E_BA⁻–H⁺ values.

2.3 SCR models of the catalysts

To disclose how the catalyst surfaces could activate the SCR, these were subjected to in situ diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy experiments with sequential N₂-balanced gases of NH₃ → NO/O₂ → NH₃ → NO/O₂ being fed to the surfaces at 220 °C, whose details are depicted in the Fig. 5 caption and the Experimental section.^11–18,33 220 °C was selected to derive the low-temperature SCR rate law for the catalysts alongside with their comparison with regard to NO_X consumption rates (−r_NOX) and kinetic parameters (E_BARRIER and lumped collision frequency ), as specified below.^11–18 The DRIFT spectra of the purged surfaces were initially recorded under a N₂ at 220 °C for their utilization as backgrounds, whose subtraction from those of the surfaces exposed to a N₂-balanced NH₃ or NO/O₂ allowed us to explore the significance of LA/BA⁻–H⁺ and ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ acting as binding spots for NH₃ and NO_X, respectively.^11–18,33

The resulting background-subtracted in situ NH₃-DRIFT spectra of the surfaces displayed multiple bands, most of which rapidly grew in positive direction and could correspond to symmetric/asymmetric stretching/wagging vibrational signals of N–H bonds pertaining to LA–NH₃ and BA⁻–NH₄⁺ species (LA and BA⁻–H⁺; Fig. 5).^{11–18,55,56} Moreover, the bands located at wavenumbers of >3500 cm⁻¹ grew simultaneously and could be indexed to asymmetric stretching vibrational signals of O–H bonds belonging to Brønsted acidic –OH species (−OH_ν,ASYM; Fig. 5), whose NH₃-mediated consumption via –O⁻–NH₄⁺ evolution led to quick growth of −OH_ν,ASYM bands in negative direction.^{11–18,57,58} Interestingly, the bands assigned to LA and −OH_ν,ASYM were as substantial as those indexed to BA⁻–H⁺. This stemmed from the domains of LA/BA⁻–H⁺/−OH_ν,ASYM bands highly intertwining from one another with their baselines being oblique markedly, as we reported elsewhere.^11–18,33 This thus made it challenging to precisely quantify the relative abundance of LA/BA⁻–H⁺/−OH species for the surfaces.^11–18,33 Nevertheless, the surfaces were found to be LA-deficient and BA⁻–H⁺-rich, as corroborated by their analytic results via CO-pulsed chemisorption, NH₃-TPD, ³¹P MAS NMR/XPS spectroscopy, etc. These could be combined with in situ NH₃-DRIFT spectral results to permit the claim that the surface BA⁻–H⁺ species had high affinity for NH₃ at low temperatures.

The component present in a N₂-balanced gas was then altered from NH₃ to NO/O₂ at 220 °C, upon which LA/BA⁻–H⁺/−OH_ν,ASYM bands were consumed rapidly in conjunction with the quick emergence of the bands being positioned at wavenumbers of 1550–1650 cm⁻¹. These bands could correspond to asymmetric stretching vibrational signals of N–O bonds of NO coordinated to ⁻O_L–M_I⁽ⁿ⁻¹⁾⁺ (NO_2ν,ASYM⁻; Fig. 5).^12,18,59,60 The intensities of NO_2ν,ASYM⁻ bands, however, were minute compared to those of LA, BA⁻–H⁺, or −OH_ν,ASYM bands, which indicated that the surfaces could hardly bind with NO. All the traits found in in situ NH₃- or NO/O₂-DRIFT spectra of the surfaces were invariant even with the continuous substitution of a N₂-balanced gas for NO/O₂ → NH₃ and NH₃ → NO/O₂ (Fig. 5). In situ DRIFT spectral results proved that the surfaces abundant with BA⁻–H⁺ species could activate the SCR via ER model, which is featured by the transition of gaseous NO to N₂ with the prime utilization of NH₃ bound to BA⁻–H⁺ (BA⁻–NH₄⁺).^11–18,33

2.4 SCR kinetics of the catalysts

BA⁻–H⁺-mediated ER model is depicted by five elementary steps, four of which are rapid and under quasi-equilibrium (eqn (1), (2), (4) and (5)) except for the sluggish RDS of NO-triggered BA⁻–NH₄⁺⋯⁻O_L–M⁽ⁿ⁻¹⁾⁺ disintegration (eqn (3)), where M corresponds to Mn/V/Sb, whereas k, K, [i]₀/[i], and C_j denote forward reaction rate constant, equilibrium constant, initial concentration/concentration of surface species i, and concentration of gaseous species j (NO_X/NH₃/H₂O), respectively.^11–18

	(1)

	(2)

	(3)

	(4)

	(5)

Eqn (3) served to describe −r_NOX (eqn (6)), whose representation was altered to eqn (9) using site balances concerning Brønsted acidic (BA⁻–H⁺; eqn (7)) and labile oxygen species (O_LMⁿ⁺; eqn (8)).^11–18

−rNOX = k3[BA−–NH4+⋯−OL–M(n−1)+]CNO	(6)

[BA−–H+]0 = [BA−–H+](1 + K1CNH3 + K1K2CNH3[OLMn+])	(7)

	(8)

	(9)

Notably, −r_NOX values of the metal vanadates were verified to be almost invariant with the alteration of C_H2O or C_NH3 values, thereby permitting to eliminate all the terms containing C_H2O/C_NH3 from the denominator of eqn (9).^11–18 Moreover, −r_NOX values should hardly rely on [O_M] corresponding to 1/2C_O2 in a feed gas.^11–18 This was because of C_O2 that was in huge excess relative to C_NO/C_NH3 in addition to the immediate regeneration of O_LMⁿ⁺ via M⁽ⁿ⁻²⁾⁺ oxidation in the redox cycle.^11–18 This thus allowed for the removal of the terms including [O_M] in the denominator of eqn (9).^11–18 Eqn (9) was thus simplified to eqn (10), wherein k₃ can be represented using eqn (11) with the inclusion of collision frequency for the RDS , given Arrhenius behaviour of the catalysts showing greater −r_NOX values at higher reaction temperatures (T_REACTION), as displayed in Table S8.†^11–18

−rNO = (k3K1K2CNH3[BA−–H+]0[Mn+OL]0)CNO	(10)

	(11)

Eqn (10) and (11) could be combined to construct Arrhenius plot of ln(−r_NOX) versus 1/T_REACTION (eqn (12)), from which E_BARRIER and can be evaluated for the catalyst.^11–18

	(12)

Notably, can be greater at a larger N_BA⁻–H⁺ (∼N_NH3 ∼ [BA⁻–H⁺]₀) or a larger N_OL (∼[Mⁿ⁺O_L]₀), whereas E_BARRIER can be lower at a lower E_BA⁻–H⁺ (∼E_NH3) or a higher E_OL, as specified above.^16–18 Besides the elaboration on N_BA⁻–H⁺/E_BA⁻–H⁺ values of the catalysts (via NH₃-TPD), their N_OL/E_OL values were also examined via XP spectroscopy (Fig. S8 and Table S5†). The XP spectra of the catalysts in the O 1s domains were curve-fitted to three sub-bands with binding energies centered at 529.9 eV, 530.3–530.7 eV, and 532.0 eV.^13–15 These could be assigned to lattice O (O_β), O_L of O_LMⁿ⁺ (O_α), and O of chemisorbed H₂O , respectively, where O_α concentrations (∼N_OL) increased in the sequence of Mn₃–P (24.2%) < Mn₂–P ∼ Mn₁–P (30.2%) < Mn₁–Sb–P (34.1%).^13–15 This corresponded to the trend on N_OL versus X anticipated in the aspect of N_{V⁵⁺–O–Mn²⁺} values (N_OL↓ at X of 1/2 → 3) along with the demonstration concerning the impartation of additional O_α species, stemming from the fragmentation of Sb⁵⁺–O–Sb⁵⁺ motifs on/near Sb₂O₅ surface. Moreover, binding energies of O_α species inherent to the catalysts were elevated in the order of Mn₃–P (530.3 eV) < Mn₂–P ∼ Mn₁–P (530.5 eV) < Mn₁–Sb–P (530.7 eV), from which O_α species of Mn₁–P/Mn₂–P were identified to be more e⁻-deficient than those of Mn₃–P and therefore derived the hierarchy on their E_OL values of Mn₁–P ∼ Mn₂–P < Mn₃–P. This was in partial accordance with the trend on E_OL versus X expected in the light of N_{Mn²⁺–O–V⁵⁺} values (E_OL↑ at X of 1 → 3).^16–18 Interestingly, the inclusion of Sb₂O₅ promoters in Mn₁–P could cause e⁻ density of O_α species pertaining to the resulting Mn₁–Sb–P to be lower than that innate to Mn₁–P, tentatively leading to E_OL values of Mn₁–P > Mn₁–Sb–P. The investigation on N_OL/E_OL values of the catalysts could suggest their reliance on the kind of Mn_XV₂O_X+5 architectures, where local environments on open/defective V or Mn cations varied dynamically with the change in X values.^30–32

One of main goals in this study was to diagnose the significance of N_BA⁻–H⁺/E_BA⁻–H⁺/N_OL/E_OL values for the catalysts in directing their −r_NOX values under a wet feed gas typically consisting of excessive H₂O/O₂ (≥3.0 vol%) and NO_X/NH₃ (800 ppm) in use across all the SCR runs, unless otherwise specified.^11–18 Notably, −r_NOX is defined by the moles of NO_X consumed in a per-BA⁻–H⁺ accessible to NH₃ at 220 °C and in a per-unit time basis (eqn (S8)†) upon the consideration of how the SCR could be activated on the surfaces (BA⁻–H⁺-mediated ER model; Fig. 1B).^11–18 Moreover, −r_NOX values should be assessed at low-temperature domains (≤250 °C), where the forward reactions substantially override the corresponding reverse ones by letting NO_X conversions (X_NOX) measured be largely deviated from those equilibrated thermodynamically (100%).^11–18 Hence, −r_NOX values were evaluated using the datasets with X_NOX values of <30%.^11–18

−r_NOX values were then related with the reciprocals of gas hourly space velocities (GHSV⁻¹) of 0.2 × 10⁻⁵–1.7 × 10⁻⁵ hours or catalyst particulate sizes (d_CATALYST) of 200–300 μm or 300–425 μm at the highest T_REACTION (250 °C) for locating reaction-limited regimes with external/internal diffusional proclivities being marginal (Fig. 4B).^11–18 Apparently, –r_NOX values were irrespective of d_CATALYST values at residence times (GHSV⁻¹) of <0.3 × 10⁻⁵ hours. This indicated that kinetic SCR domains free from diffusional artifacts were within the residence times of <0.3 × 10⁻⁵ hours at ≤250 °C and therefore served throughout the other kinetic SCR runs (sky-blue-shaded rectangle in Fig. 4B).^11–18 Obviously, −r_NOX values of the catalysts were greater at higher T_REACTION values (Table S8†), which demonstrated their conformity to Arrhenius behaviour.^11–18 Furthermore, regardless of the change in T_REACTION values, −r_NOX values retained their hierarchy of Mn₃–P < Mn₂–P ∼ Mn₁–P < Mn₁–Sb–P (e.g., 3.4 × 10⁻¹ min⁻¹ for Mn₃–P; 5.4 × 10⁻¹ min⁻¹ for Mn₂–P/Mn₁–P; 6.2 × 10⁻¹ min⁻¹ for Mn₁–Sb–P at 250 °C), while revealing N₂ selectivities (S_N2) of 100%. This proved the superiority of Mn₁–Sb–P to Mn_X–P catalysts, among which the preferential choice of Mn₁–P over Mn₂–P as a reservoir of promotive Sb₂O₅ species was detailed below.

Arrhenius plots were finally constructed for the catalysts to assess their E_BARRIER and values, both of which served to unlock the centrality of N_BA⁻–H⁺/E_BA⁻–H⁺/N_OL/E_OL in dominating −r_NOX values (Fig. 4C).^11–18 Their E_BARRIER values increased in the order of Mn₁–Sb–P ∼ Mn₁–P ∼ Mn₂–P (48.8–50.7 kJ mol⁻¹) < Mn₃–P (53.4 kJ mol⁻¹), whose trend was partially con-current to that on their E_BA⁻–H⁺ values (Mn₁–Sb–P < Mn₁–P ∼ Mn₂–P < Mn₃–P), yet, was hardly linked to that on their E_OL values (Mn₁–Sb–P < Mn₁–P ∼ Mn₂–P < Mn₃–P) considering the relationships of E_BA⁻–H⁺↓→E_BARRIER↓ and E_OL↑→E_BARRIER↓, as discussed earlier.^16–18 Meanwhile, values were elevated in the sequence of Mn₁–P ∼ Mn₂–P (4.2 × 10⁴ min⁻¹) < Mn₃–P ∼ Mn₁–Sb–P (7.1 × 10⁴ min⁻¹), whose hierarchy was in accordance with that on their N_BA⁻–H⁺ values and did not match that on their N_OL values (Mn₃–P < Mn₂–P ∼ Mn₁–P < Mn₁–Sb–P) given the correlations of , as depicted in the Introduction.

Indeed, E_BA⁻–H⁺/N_BA⁻–H⁺ overrode E_OL/N_OL in directing −r_NOX values of the catalysts. This demonstrated that the acidic cycle could outweigh the redox counterpart in accelerating the SCR under reaction-limited regions, even with the association of NH₃-chemisorbed Brønsted acidic (BA⁻–NH₄⁺) and labile oxygen species (⁻O_L–M⁽ⁿ⁻¹⁾⁺) to form BA⁻–NH₄⁺⋯⁻O_L–M⁽ⁿ⁻¹⁾⁺ (eqn (3) and Fig. 1B).^11–18 In this sense, it was scientifically convincing that −r_NOX could be an indicative of overall acidic cycling efficiency for the catalyst. Importantly, kinetic SCR runs validated the prime descriptor of −r_NOX values was the X values of Mn_XV₂O_X+5 architectures, whose local Mn/V environments considerably affected the hierarchies on E_BA⁻–H⁺/N_BA⁻–H⁺ values for the catalysts.

2.5 Redox properties of the catalysts

Although the redox cycle barely contributes to inducing −r_NOX values under reaction-limited domains, it is as central as acidic counterpart under diffusion-limited regions, where all the surface features are combined to reveal X_NOX/S_N2 values maximum-achievable at T_REACTION of ≤400 °C.^11–18 The redox traits of the catalysts were attempted to be examined by their exposure to a H₂ at 50–800 °C at β value of 10 °C min⁻¹, from which their H₂-temperature-programmed reduction (H₂-TPR) profiles (TCD signal versus temperature) were monitored, yet, were of only a little difference (Fig. S9†).^11–18 Exception could be found in the areas under the H₂-TPR profiles of the catalysts, which could correspond to the amounts of defective Mn/V/Sb cations and BA⁻–H⁺/O_M/O_L liable to H₂ reduction in a per-gram basis (N_H2) with the hierarchy of Mn₁–P ∼ Mn₂–P ∼ Mn₃–P (0.6–0.8 mmol_H2 g_CAT⁻¹) < Mn₁–Sb–P (1.2 mmol_H2 g_CAT⁻¹). However, thermal energies (temperatures) needed to reduce the individual surface species of the catalysts highly overlapped from one another, thereby incurring their H₂-TPR profiles to provide little insights into the surface species involved in the redox cycle.

Meanwhile, the redox cycle is activated by O_L of O_LMⁿ⁺ and O_M confined on O_V, whose hierarchies concerning N_OM and E_OM values expected for the catalysts were Mn₃–P < Mn₂–P ∼ Mn₁–P and Mn₁–P ∼ Mn₂–P < Mn₃–P, respectively, in the light of N_{V⁵⁺–O–Mn²⁺} values, as opposed to the trend on their E_OM values of Mn₁–P > Mn₂–P > Mn₃–P in the aspect of N_{Mn²⁺–O–V⁵⁺} values.^15–18 In this regard, O₂-pulsed chemisorption technique was utilized to evaluate the numbers of O₂ (∼2O_M) chemisorbed in a per-gram of the catalysts (N_O2), whose O_V sites available to bear O_M species at low temperatures were void via H₂ reduction at 300 °C prior to chemisorbing O₂ molecules at 250 °C (Table S4†).^15–18 The resulting N_O2 values increased in the order of Mn₃–P (21.5 μmol_O2 g_CAT⁻¹) < Mn₂–P ∼ Mn₁–P ∼ Mn₁–Sb–P (27.2–29.5 μmol_O2 g_CAT⁻¹). This indicated the local environments on defective Mn cations were more crucial to dictate N_OM values of the catalysts than those on open/defective V counterparts. Moreover, the similarity found in N_O2 values of Mn₁–P and Mn₁–Sb–P indicated that defective Sb cations on fragmented Sb⁵⁺–O–Sb⁵⁺ channels of Sb₂O₅ promoters could prefer binding with P⁵⁺–O²⁻ bonds of PO₄³⁻ functionalities (BA⁻) rather than coordinating with O_M species.

O₂-TPD experiments were conducted for the catalysts subjected to initial H₂ reduction at 300 °C, upon which O_M species (1/2O₂) were chemisorbed on vacant O_V sites at 250 °C prior to their release from O_V sites with the surface temperatures being elevated from 250 °C to 700 °C at β value of 10 °C min⁻¹.^15–18 The areas under the resulting O₂-TPD profiles (TCD signal versus temperature) of the catalysts then served to compare their N_OM values, yet, were not substantially dissimilar from one another (not shown). Additional O₂-TPD runs were thus performed identically to those depicted above except for the change in O₂ chemisorption temperature from 250 °C to 50 °C with diverse β values of 10–30 °C min⁻¹ (Fig. S10, S11 and Table S9†).^15–18 Again, N_OM values of the catalysts were assessed using the areas under their O₂-TPD profiles relative to that of Mn₁–P set as 1.0, whose hierarchy was Mn₃–P (0.7) < Mn₂–P ∼ Mn₁–P ∼ Mn₁–Sb–P (0.9–1.1), as shown in Fig. 6A. The trend found in N_OM values of the catalysts was identical to that found in their N_O2 values, which again could prove defective Mn cations of Mn_XV₂O_X+5 building units pivotal to direct N_OM/N_O2 values of the catalysts in conjunction with fragmented Sb⁵⁺–O–Sb⁵⁺ motifs on Mn₁–Sb–P surface to hardly bear O_M species. O₂-TPD profiles were de-convoluted into three-sub-bands, whose T_MAX values served to construct the plots of ln(β/T_MAX²) versus 1/T_MAX for evaluating the slopes corresponding to −E_OM/R values (Fig. 6A and Table S9†).^15–18 E_OM values were elevated in the sequence of Mn₁–Sb–P (22.1 kJ mol_O2⁻¹) < Mn₁–P ∼ Mn₂–P (25.6 kJ mol_O2⁻¹) < Mn₃–P (29.1 kJ mol_O2⁻¹), which matched the trend on E_OM versus X predicted in the light of N_{V⁵⁺–O–Mn²⁺} values.^16–18 This further substantiated that defective Mn cations could function as the prime dictator of O_M properties (N_OM/E_OM) for the catalysts. Moreover, Mn₁–P bore defective Mn cations inclined to e⁻ acceptance from Sb⁵⁺ components of Sb₂O₅ promoters, thus rendering O_V sites adjacent to defective Mn cations of the resulting Mn₁–Sb–P to be less electrophilic than those of Mn₁–P, leading to E_OM values of Mn₁–Sb–P < Mn₁–P, as posited earlier. Importantly, given the hierarchy on E_OM values of the catalysts, their O_M mobilities were expected to increase in the sequence of Mn₃–P < Mn₂–P ∼ Mn₁–P < Mn₁–Sb–P.^16–18

The trend on E_OM values (O_M mobilities) of the catalysts were coupled with those on their N_OM/N_OL values to induce the tentative rank on the overall redox cycling efficiencies for the catalysts of Mn₃–P < Mn₂–P ∼ Mn₁–P < Mn₁–Sb–P. This was tested by exposing the catalysts under diffusion-limited time-on-stream SCR conditions, which were featured by a low temperature (250 °C) and a long residence time (GHSV⁻¹ of 1.7 × 10⁻⁵ hours; space velocity of 60000 h⁻¹) in tandem with the exclusion of O₂ from a wet feed gas at 2–8 hours (denoted as O₂-on/off run; Fig. 6B).^11–18 In addition, the kind of O₂ (O_M precursor) used for O₂-on/off runs was altered from a generic ¹⁶O₂ to a heavier, labelled ¹⁸O₂, whose purpose was to rigorously compare the overall redox cycling efficiencies of the catalysts, as gauged by their X_NOX values with ¹⁸O₂ being turned-off relative to those with ¹⁸O₂ being turned-on (X_NOX/X_NOX,0).^15–18 Notably, S_N2 values of the catalysts were 100% throughout all the SCR runs under diffusion-limited domains (including O₂-on/off runs), thereby being omitted from the discussion. Initially, the catalyst surfaces were reduced by H₂ at 300 °C to vacate all the O_V sites available to bear ¹⁸O_M species.^15–18 The surfaces were then in contact with ¹⁸O₂ for saturating their O_V sites with ¹⁸O_M species at 250 °C and subjected to the SCR runs under a wet feed gas containing ¹⁸O₂ for monitoring their X_NOX/X_NOX,0 values at 250 °C.^15–18 The resulting X_NOX/X_NOX,0 values of the catalysts were 1.0 with ¹⁸O₂ being turned-on up to 2 hours, plummeted as soon as ¹⁸O₂ was turned-off at ≥2 hours, plateaued at ≤8 hours, yet, were instantaneously recovered to 1.0 with ¹⁸O₂ being re-included in a feed gas. This indicated that X_NOX/X_NOX,0 values of the catalysts were heavily dependent on their abilities to accelerate the redox cycle via the recuperation of ¹⁸O_M and ¹⁸O_L via ¹⁸O₂ supply and ¹⁸O_M migration, respectively, during O₂-on/off runs.^15–18 Of importance was that the decrease in X_NOX/X_NOX,0 values were more pronounced in the sequence of Mn₁–Sb–P < Mn₁–P ∼ Mn₂–P < Mn₃–P with ¹⁸O₂ being excluded from a feed gas at 2–8 hours. This apparently demonstrated that the overall redox cycling efficiencies of the catalysts were higher in the order of Mn₃–P < Mn₂–P ∼ Mn₁–P ∼ Mn₁–Sb–P, which was in agreement with those anticipated using their N_OL/N_OM/E_OM values.

2.6 H₂O resistance and SCR performance of the catalysts

Hydrophobicity of the catalysts can dictate their X_NOX/S_N2 values maximum-obtainable under diffusion-limited domains, where O²⁻ of H₂O can bind with H⁺ of BA⁻–H⁺ through the formation of BA⁻⋯H₃O⁺ with the relationship of E_BA⁻–H⁺↑ → H₂O binding energy (E_H2O)↑ or coordinate to O_V sites vicinal to defective Mn/V/Sb cations with the relationship of E_OM↑ → E_H2O↑.^16–18,61 H₂O adsorption isotherms of the catalysts were thereby collected at 0–40 °C, across which the numbers of H₂O adsorbed on the surfaces versus relative H₂O pressure were monitored up to the points free from the instrumental constraints (Fig. S12 and Table S10†).^16–18 The resulting H₂O adsorption isotherms were simulated via Toth fitting to evaluate their isosteric heats of H₂O adsorption (E_H2O) at near-zero H₂O coverages via Clausius–Clapeyron equation (Fig. 7A).^16–18 E_H2O values were elevated in the order of Mn₁–Sb–P (21.2 kJ mol_H2O⁻¹) < Mn₁–P ∼ Mn₂–P (27.5 kJ mol_H2O⁻¹) < Mn₃–P (30.5 kJ mol_H2O⁻¹), which could match and re-affirm the trends found in their E_NH3/E_OM values (Mn₁–Sb–P < Mn₁–P ∼ Mn₂–P < Mn₃–P). Moreover, H₂O-accessible BET surface areas in a per-gram basis for the catalysts (S_BET,H2O) showed the hierarchy of Mn₁–P < Mn₂–P ≤ Mn₃–P at 0–40 °C (e.g., 38.0 m_H2O² g_CAT⁻¹ for Mn₁–P; 82.2 m_H2O² g_CAT⁻¹ for Mn₂–P/Mn₃–P at 25 °C). This could corroborate the superiority of Mn₁–P to Mn₂–P in resisting H₂O, thus rationalizing the preferential selection of the former over the latter for improvement with the use of promotive Sb₂O₅ species, in spite of S_BET,H2O values of Mn₁–P < Mn₁–Sb–P (64.5 m_H2O² g_CAT⁻¹), probably resulting from additional H₂O binding spots (BA⁻–H⁺) on Sb₂O₅ surface.

The overall H₂O tolerance of the catalysts was investigated via their time-on-stream SCR runs under diffusion-limited, low-temperature regimes (space velocity of 60000 h⁻¹; 220 °C) in conjunction with the inclusion of H₂O in an O₂-plentiful feed gas at 1–4 hours (denoted as H₂O-on/off run; Fig. 7B).^16–18 X_NOX values of the catalysts with H₂O turned-on relative to those with H₂O turned-off (X_NOX/X_NOX,0) served to rank the capabilities of their BA⁻–H⁺/O_M species reluctant to binding with H₂O (X_NOX/X_NOX,0↑ → overall H₂O resistance↑).^16–18 Indeed, X_NOX/X_NOX,0 values were elevated in the sequence of Mn₃–P (0.77) < Mn₂–P ∼ Mn₁–P (0.83) < Mn₁–Sb–P (0.94) with H₂O being turned-on at 1–4 hours, which matched those anticipated for the catalysts, as established based on their E_H2O and/or S_BET,H2O values.

The inspection on the overall acidic cycling (−r_NOX), overall redox cycling (O₂-on/off), and hydrophobic characteristics (H₂O-on/off) for the catalysts could allow to postulate that their overall SCR efficiencies were higher in the sequence of Mn₃–P < Mn₂–P ∼ Mn₁–P < Mn₁–Sb–P. To test this postulation, the catalysts were exposed to an O₂-ample, wet feed gas under diffusion-limited regions (space velocity of 60000 h⁻¹) with T_REACTION values being altered from 150 to 400 °C (denoted as temperature-sweep run; Fig. 7C).^11–18,33 Indeed, the catalysts elevated their X_NOX values maximum-achievable in the sequence of Mn₃–P < Mn₂–P ∼ Mn₁–P < Mn₁–Sb–P at <250 °C. In particular, Mn₁–P improved the overall SCR efficiencies (X_NOX) of ∼15.0% upon the inclusion of Sb₂O₅ promoters at 200–220 °C, leading the resulting Mn₁–Sb–P to exhibit the highest SCR performance among all the catalysts investigated herein.

Moreover, the overall SCR efficiency of Mn₁–Sb–P was compared with that of Mn₁–Sb–S, for which Mn₁–Sb host was functionalized with SO_A²⁻ guests at the identical temperature (500 °C) to that utilized to synthesize Mn₁–Sb–P for fair comparison.¹⁴ 500 °C was also chosen to impart BA⁻ species of the resulting Mn₁–Sb–S with substantial hydrophobicity conducive to promote ABS pyrolysis efficiency, as we demonstrated previously.¹⁴ Of note was that SO_A²⁻ binding modes of Mn₁–Sb–S were mostly bi-dentate, as verified with the use of in situ SO₂/O₂-DRIFT spectroscopy experiment of Mn₁–Sb at 500 °C.¹⁴ Of additional note was that Mn₁–Sb–P and Mn₁–Sb–S were comparable with regard to the molar ratios of heteroatom relative to metal in the bulk/surface scales (P/(Mn + V + Sb) of 0.2 (±0.1)/0.7 (±0.1) for Mn₁–Sb–P; S/(Mn + V + Sb) of 0.2 (±0.1)/0.5 (±0.2) for Mn₁–Sb–S).¹⁴ These were gathered to induce the expectation that N_NH3 (∼N_BA⁻–H⁺) of the former could be greater than that of the latter at low temperatures. NH₃-TPD experiment of Mn₁–Sb–S was thereby conducted to assess its N_NH3 (105 μmol_NH3 g_CAT⁻¹) at 220 °C and corroborated N_NH3 (∼N_BA⁻–H⁺) values of Mn₁–Sb–S < Mn₁–Sb–P (120 μmol_NH3 g_CAT⁻¹) at 220 °C (Fig. 4A and S13A†). The overall SCR efficiencies (X_NOX) of Mn₁–Sb–S were 13–23% lower than those of Mn₁–Sb–P at T_REACTION values of 180–220 °C (Fig. 7C). This could be owing to the inferiority of the former to the latter in elevating the , overall redox cycling trait, or overall hydrophobic feature at low temperatures. This could demonstrate the merit of PO₄³⁻ functionality over SO_A²⁻ counterpart functioning as a conjugate base of BA⁻–H⁺ or a modifier of the redox site properties under an O₂-abundant, wet feed gas.

Furthermore, the overall SCR efficiency of Mn₁–Sb–P was also compared with that of a commercial control with bulk V content of 2.0 wt% (V₂O₅–WO₃), whose surface, yet, was not functionalized with PO₄³⁻ modifiers. This was because of previous SCR reports showing the decrease in low-temperature SCR performance for V₂O₅–WO₃ subjected to oxidative phosphorus modification (bulk P contents of ≥0.2 wt% for those reported; 0.4 wt% for Mn₁–Sb–P).^62–64 The overall SCR efficiencies (X_NOX) of Mn₁–Sb–P and V₂O₅–WO₃ were comparable across all the T_REACTION values being dialled-in. Mn₁–Sb–P and V₂O₅–WO₃ were thereby subjected to hydro-thermal aging (HT) via their exposure to an H₂O/O₂-abundant N₂ at 600 °C for 150 hours for additional X_NOX comparison of the resulting Mn₁–Sb–P-HT and V₂O₅–WO₃-HT. Mn₁–Sb–P barely lost its overall SCR efficiency even post undergoing HT, as gauged by X_NOX reduction of ≤∼5.0% for Mn₁–Sb–P-HT relative to Mn₁–Sb–P at all the T_REACTION values. This was in contrast to V₂O₅–WO₃-HT, whose X_NOX values were 15–50% smaller than those of V₂O₅–WO₃ at 180–250 °C. This could be due in part to weaker H₂O tolerance of V₂O₅–WO₃ in comparison with that of Mn₁–Sb–P, as substantiated by their X_NOX/X_NOX,0 values monitored during H₂O-on/off runs at 220 °C (Fig. 7B).^16–18 X_NOX/X_NOX,0 values of V₂O₅–WO₃ and Mn₁–Sb–P were 0.88 and 0.94, respectively, with H₂O being turned-on at 1–4 hours. The temperature-sweep and H₂O-on/off runs demonstrated the superiority of Mn₁–Sb–P to Mn₁–Sb–S or V₂O₅–WO₃ in activating the low-temperature SCR under an O₂-rich wet feed gas.

2.7 Poison accumulation/pyrolysis on the catalysts

Mn₁–Sb–P and V₂O₅–WO₃ were further examined to disclose which one outperformed the other in tolerating AS/ABS poisons upon the inclusion of SO₂ in a wet feed gas. AS/ABS tolerance of the catalysts stated above was also occasionally compared with that of Mn₁–Sb–S using its datasets we reported elsewhere or acquired herein due to the lowest overall SCR efficiencies of Mn₁–Sb–S maximum-achievable at ≤400 °C (Fig. 7C) and/or dissimilar environments utilized to poison the catalyst surfaces with AS/ABS.¹⁴ SO₂ can interact with O_L/O_M to form SO₃ on/near the catalyst surface.^65–67 SO₃ is prone to bind with NH₃ and H₂O in proximity, resulting in the production of surface SO₃⋯H₂O⋯NH₃ and SO₃⋯(H₂O)₂⋯NH₃ intermediates essential to trigger the deposition of AS ((NH₄)₂SO₄) and ABS ((NH₄)HSO₄) poisons on the surface, respectively.^{12–14,16–18,68,69} AS/ABS incurs NH₃ and 1/2O₂ (O_M precursor) to be in limited access to BA⁻–H⁺ and O_V sites, respectively, and thereby persistently hinder the activation of the acidic/redox cycles for the SCR, unless being dissociated at <300 °C.^14,16–18 AS/ABS productivity can increase with a larger number of SO₂ chemisorbed in a per-gram of the catalyst (N_SO2) and its higher binding intimacy with SO₃ (E_SO3) alongside with its higher N_NH3, E_NH3, S_BET,H2O, E_H2O, etc., among which S_BET,H2O and/or E_H2O could be higher in V₂O₅–WO₃ than in Mn₁–Sb–P.^14,16–18 This was partially demonstrated by their H₂O-on/off runs with X_NOX/X_NOX,0 values of the former being lower than those of the latter when H₂O was contained in a feed gas (Fig. 7B).

In this regard, Mn₁–Sb–P/V₂O₅–WO₃ were subjected to SO₂-TPD experiments, where the surfaces chemisorbed SO₂ molecules at 220 °C prior to being heated to 900 °C with diverse β values of 10–20 °C min⁻¹ (Fig. S14 and Table S11†).^14–18 The areas under the resulting SO₂-TPD profiles (SO₂ concentration (C_SO2) versus temperature) of the catalysts served to quantify their N_SO2 values, from which N_SO2 of V₂O₅–WO₃ was around 2-fold greater than that of Mn₁–Sb–P (32.2 μmol_SO2 g_CAT⁻¹), as shown in Fig. 8A.^14–18 Moreover, TPD theory served to plot ln (β/T_MAX²) versus 1/T_MAX for individual sub-band I–V, stemming from the SO₂-TPD profiles of the catalysts subjected to de-convolution (Fig. 8A and Table S11†).^14–18 These plots allowed to evaluate −E_SO2/R values for V₂O₅–WO₃ and Mn₁–Sb–P, from which E_SO2 of the former was identified to be around 1.5-fold higher than that of the latter.^14–18 This suggested that SO₃ binding affinity for the surface might be greater in V₂O₅–WO₃ than in Mn₁–Sb–P, as speculated based on their E_SO2 values of V₂O₅–WO₃ > Mn₁–Sb–P. Furthermore, V₂O₅–WO₃ was subjected to NH₃-TPD experiment, whose protocols were identical to those of NH₃-TPD experiment on Mn₁–Sb–P with NH₃ molecules being chemisorbed at 220 °C along with the surface temperature elevation to 700 °C with β value of 10 °C min⁻¹ (Fig. S13B†).^11–18,33 The area under the resulting NH₃-TPD profile served to evaluate N_NH3 of V₂O₅–WO₃ at 220 °C, where N_NH3 of V₂O₅–WO₃ (220 μmol_NH3 g_CAT⁻¹) was approximately twice that of Mn₁–Sb–P. In sum, in spite of lacking E_NH3 of V₂O₅–WO₃ at 220 °C, its N_SO2, E_SO2, N_NH3, and E_H2O (hydrophilicity) values were greater than the corresponding values of Mn₁–Sb–P. This was deemed sufficient for constructing the hypothesis on a greater amount of AS/ABS deposited on V₂O₅–WO₃ surface compared to that on Mn₁–Sb–P counterpart upon their exposure to a SO₂-bearing, wet feed gas at <300 °C. To test this hypothesis, the catalyst surfaces were exposed to a wet feed gas containing SO₂ at 180 °C for 30 hours to accumulate AS/ABS poisons.^14,16–18 The resulting AS/ABS-poisoned catalysts were then underwent thermogravimetric analyzer (TGA)-mass spectrometer (MASS) experiments (Fig. S15 and Table S12†), during which the poisoned surfaces were initially purged to remove physisorbed H₂O prior to being heated to 600 °C under an Ar with β value of 5 °C min⁻¹.^14,16–18 The resulting TGA profiles (weight percent (wt%; W) loss versus temperature) permitted us to evaluate the quantities of AS/ABS pertaining to the poisoned catalysts (N_AS/ABS) using their W loss values (ΔW) at ≤500 °C, thus ruling out the chance concerning pyrolysis of PO₄³⁻ functionalities grafted on Mn₁–Sb–P surface at >500 °C.^14,16–18 Indeed, N_AS/ABS (∼ΔW) of Mn₁–Sb–P (6.9 wt%) was smaller than that of V₂O₅–WO₃ (11.6 wt%), which could elevate the likelihood concerning a higher AS/ABS resistance for the former than that for the latter (Fig. 8A). In our previous study, Mn₁–Sb–S was poisoned with AS/ABS under the identical conditions to those used to produce AS/ABS-poisoned Mn₁–Sb–P/V₂O₅–WO₃ except for the utilization of a higher C_H2O (7.7 vol%) than that used herein (5.4 vol% H₂O).¹⁴ A higher C_H2O could generate a larger amount of AS/ABS precursors (SO₃⋯H₂O⋯NH₃/SO₃⋯(H₂O)₂⋯NH₃) on Mn₁–Sb–S surface, whose reported N_AS/ABS (∼ΔW; 15.8 wt%; Fig. S16†) could thus hardly serve for fair comparison with those of Mn₁–Sb–P/V₂O₅–WO₃.^{12–14,16–18,68,69}

Meanwhile, AS can be deaminated for its transformation into ABS at <260 °C.^14,16–18 ABS can then be thermally decomposed into SO₂, O₂, NH₃, H₂O, and N₂ by way of a series of pathways, all of which are thermodynamically spontaneous.^14,16–18 Specifically, ABS can be dehydrated to produce ammonium pyrosulfate (1/2(NH₄)₂S₂O₇) consisting of NH₄⁺ and 1/2S₂O₇²⁻ fragments, where NH₄⁺ can bind with ⁻O_L–M⁽ⁿ⁻¹⁾⁺ to transform into NH₃⋯H⁺–⁻O_L–M⁽ⁿ⁻¹⁾⁺, which in turn can lose 1/6N₂, 1/2H₂, and 2/3NH₃ to yield HO_L–M⁽ⁿ⁻¹⁾⁺ prior to the recovery of ⁻O_L–M⁽ⁿ⁻¹⁾⁺ via dehydration/oxidation of HO_L–M⁽ⁿ⁻¹⁾⁺ (denoted as NH₄⁺ fragmentation cycle I) with the transition of HO_L–M⁽ⁿ⁻¹⁾⁺ → ⁻O_L–M⁽ⁿ⁻¹⁾⁺ being depicted in Fig. 1B, eqn (4) and (5).^14,16–18 Interestingly, NH₃⋯H⁺–⁻O_L–M⁽ⁿ⁻¹⁾⁺ can also be regarded as NH₄⁺–⁻BA (BA⁻–NH₄⁺) used to activate the acidic cycle of the BA⁻–H⁺-mediated ER model (referred to as NH₄⁺ fragmentation cycle II; Fig. 1B).^14,16–18 The NH₄⁺ fragmentation cycle I/II were verified to be readily activated and thermodynamically favorable, while demanding lower E_BARRIER values in comparison with that of 1/2S₂O₇²⁻ fragmentation cycle detailed in Fig. 8B, where 1/2 of 1/2S₂O₇²⁻ are omitted for simplicity.^14,16–18

S₂O₇²⁻ can undergo transition from 2(BA⁻–H⁺⋯⁻O–SO₂)O, 2(BA⁻–H⁺⋯⁻O–SO₂–O⁻⋯H⁺), to 2(BA⁻⋯H₂O–SO₂–H₂O) via interplays with 2BA⁻–H⁺, H₂O chemisorbed on the surface, and 2H₂ wandering in the surface, respectively (Fig. 8B).^14,16–18 H₂O and H₂ can be in excess relative to 2(BA⁻–H⁺⋯⁻O–SO₂)O and 2(BA⁻–H⁺⋯⁻O–SO₂–O⁻⋯H⁺), respectively, thereby accelerating 2(BA⁻–H⁺⋯⁻O–SO₂)O + H₂O → 2(BA⁻–H⁺⋯⁻O–SO₂–O⁻⋯H⁺) and 2(BA⁻–H⁺⋯⁻O–SO₂–O⁻⋯H⁺) + 2H₂ → 2(BA⁻⋯H₂O–SO₂–H₂O) in rapid and exothermic manner (Fig. 8B).^14,16–18 This is in contrast to the rate-directing, endothermic disintegration of 2(BA⁻⋯H₂O–SO₂–H₂O) to liberate 4H₂O/2SO₂ with 2BA⁻ being left on the surface yet highly prone to protonation mediated by ample H⁺ roaming in the surface (*; Fig. 8B).^14,16–18 Of significance was that E_BARRIER required to activate 2(BA⁻⋯H₂O–SO₂–H₂O) → 4H₂O + 2SO₂ + 2BA⁻ can be lower if the binding affinity of BA⁻ for H⁺ of H₂O belonging to H₂O–SO₂–H₂O (denoted as E_BA⁻) is smaller.^14,16–18 Of additional significance was that H₂O-on/off runs on V₂O₅–WO₃/Mn₁–Sb–P could allow for evaluating their (relative) E_H2O values (Fig. 7B). The E_H2O values stated above, however, can be defined by the binding energies between H⁺ of protonated BA⁻(BA⁻–H⁺) and O²⁻ of H₂O, from which a higher E_H2O can induce a greater tendency to generate BA⁻⋯H₃O⁺ (not BA⁻⋯H₂O) via an easier H⁺ donation from BA⁻–H⁺ to H₂O.^17,61 Hence, it could be challenging to predict which one possessed a higher E_BA^– than the other on the ground of their H₂O-on/off runs. On the other hand, low-temperature S₂O₇²⁻ fragmentation cycle on the catalysts can be mainly activated by their BA⁻–H⁺ species, whose quantities correspond to their N_NH3 values functioning as the dictators of collision frequencies between BA⁻–H⁺ and S₂O₇²⁻, between 2(BA⁻–H⁺⋯⁻O–SO₂)O and H₂O, or between BA⁻–H⁺⋯⁻O–SO₂–O⁻⋯H⁺ and H₂ (Fig. 8B).^14,16–18 Hence, a higher was expected to be achievable in V₂O₅–WO₃ rather than in Mn₁–Sb–P due to a larger N_NH3 provided by the former compared to that imparted by the latter.

AS/ABS pyrolysis efficiencies of the catalysts were compared using their AS/ABS consumption rates (−r_AS/ABS), which are defined by the moles of AS/ABS consumed in a per-BA⁻–H⁺ accessible to NH₃ at 220 °C and in a per-unit time basis.^14,16–18 −r_AS/ABS values of the catalysts could be assessed using their TGA-MASS datasets to provide ΔW values within a time span (Δt) with the deviation of ±0.5 °C from a T_REACTION being set (Fig. S15A, B and Table S12†).^14,16–18 Again, AS can be converted readily to ABS under an Ar at <260 °C.^14,16–18 This led us to consider kinetic datasets of the catalysts at 260–290 °C for making their −r_AS/ABS values be identical to −r_ABS analogues (eqn S(9)†).^14,16–18 For rigor, −r_ABS values of the catalysts were evaluated using kinetic datasets with ABS conversions (X_ABS) of ≤20.0% only, whose magnitude was far away from that thermodynamically equilibrated (100%) at 260–290 °C.^14,16–18 For additional rigor, the RDS of S₂O₇²⁻ fragmentation cycle can be activated to release SO₂, whose signals emitted versus temperature for the catalysts were also recorded during their TGA-MASS experiments (Fig. S15C and D†).^14,16–18 The onsets of SO₂ signal evolution (T_ONSET) were found to be 226 °C and 239 °C for Mn₁–Sb–P and V₂O₅–WO₃, respectively, which could add weigh on the claim that −r_ABS values of the catalysts should be inspected at 260–290 °C for accurately assessing their E_BARRIER values. It should be stressed that a smaller E_BARRIER was anticipated to be achievable in Mn₁–Sb–P than in V₂O₅–WO₃ because T_ONSET could be an indicative of the thermal energy needed for the RDS activation, in addition to T_ONSET values for the catalysts of Mn₁–Sb–P < V₂O₅–WO₃.^14,16–18 Interestingly, in spite of a higher C_H2O in use for poisoning Mn₁–Sb–S with AS/ABS, T_ONSET reported for Mn₁–Sb–S (237 °C; Fig. S16†) was as low as that for V₂O₅–WO₃, which suggested that E_BARRIER values for Mn₁–Sb–S and V₂O₅–WO₃ could be in similar magnitude.¹⁴

−r_ABS values of Mn₁–Sb–P were consistently higher than those of V₂O₅–WO₃ at all the T_REACTION values tested (Table S12†), which indicated that the former was superior to the latter in catalyzing ABS pyrolysis, tentatively resulting from a smaller E_BARRIER provided by the former than that imparted by the latter. Notably, a higher C_H2O previously used to poison Mn₁–Sb–S with AS/ABS could be coupled with its N_NH3 to markedly affect of Mn₁–Sb–S along with its −r_ABS values in a per-BA⁻–H⁺ site basis at ≥260 °C, thus rendering the fair comparison of −r_ABS values for Mn₁–Sb–S versus the others to be of challenge.^14,16–18

Nevertheless, −r_ABS values of the catalysts were greater at higher T_REACTION values, indicating the catalysts obeyed Arrhenius behavior (Fig. S16 and Table S12†).^14,16–18 −r_ABS value can be represented using the generic rate law (eqn (13)), where k_APP, k_APP,0, C_ABS, and l are referred to as apparent ABS consumption (reaction) rate constant, collision frequency, ABS concentration, and reaction order, respectively.^14,16–18

	(13)

Arrhenius plots of ln(−r_ABS) versus 1/T_REACTION were constructed for the catalysts (eqn (14)) to evaluate their E_BARRIER values and lumped collision frequencies , as detailed in Fig. 8C and S16.†^14,16–18

	(14)

Apparently, of V₂O₅–WO₃ was verified to be around 3-fold higher than that of Mn₁–Sb–P (0.3 × 10² min⁻¹), which was in agreement with the trend on their N_NH3 values of V₂O₅–WO₃ > Mn₁–Sb–P, as expected earlier. Conversely, E_BARRIER of V₂O₅–WO₃ was found to be around 6.5 kJ mol⁻¹ higher than that of Mn₁–Sb–P (26.5 kJ mol⁻¹). This was in accordance with their T_ONSET values of V₂O₅–WO₃ > Mn₁–Sb–P. This also corroborated that the hydrophobicity of BA⁻ species was greater in Mn₁–Sb–P than in V₂O₅–WO₃, leading the former to reduce E_BARRIER necessary for ABS dissociation (RDS activation) more readily than the latter.

On the whole, AS/ABS pyrolysis efficiencies of Mn₁–Sb–P > V₂O₅–WO₃ could prove the claim that hydrophobicity of BA⁻ species can override the amount of their protonated analogues (BA⁻–H⁺) in dictating −r_ABS under a low thermal energy. In addition, −r_ABS values of the catalysts were coupled with their N_AS/ABS values to allow for the postulation that Mn₁–Sb–P can enhance AS/ABS resistance over V₂O₅–WO₃ at low temperatures. Moreover, in contrary to of Mn₁–Sb–S, its E_BARRIER should not rely on either C_H2O or N_NH3 because −r_ABS values were evaluated in a per-BA⁻–H⁺ site basis.^14,16–18 However, E_BARRIER of Mn₁–Sb–S (30.0 (±2.7) kJ mol⁻¹; Fig. S16†) was comparable to that of V₂O₅–WO₃ (32.9 (±2.9) kJ mol⁻¹) and therefore posing the need to compare their AS/ABS resistance via SCR runs under regulated conditions.

In this regard, the catalysts were subjected to time-on-stream SCR runs (denoted as SO₂-on/off runs; Fig. 9A), whose environments were identical to those used to conduct H₂O-on/off runs expect for the consistent supply of an O₂-ample, wet feed gas into the catalyst surfaces in tandem with the addition of SO₂ to the resulting O₂/H₂O-containing feed gas at ≥2 hours.^11–18,33 X_NOX values of the catalysts with SO₂ turned-on relative to those with SO₂ turned-off (X_NOX/X_NOX,0) served to gauge AS/ABS tolerance of the catalysts at ≥2 hours by monitoring their time spans need to reach X_NOX/X_NOX,0 values of 0.7 (denoted as t_0.7).^11–18,33 t_0.7 of Mn₁–Sb–P was approximately 2.5-fold longer than that of V₂O₅–WO₃ (8 hours), again originating from N_AS/ABS values of Mn₁–Sb–P < V₂O₅–WO₃ and −r_ABS values of Mn₁–Sb–P > V₂O₅–WO₃. Interestingly, X_NOX/X_NOX,0 values of Mn₁–Sb–S gradually increased from 1.0 to 1.25 at 2–5 hours. This could originate from the generation of additional SO_A²⁻ (BA⁻) modifiers on Mn₁–Sb–S surface and the subsequent expedition of its acidic cycle via protonation of the corresponding BA⁻ species at 2–5 hours, as also found in our previous studies on metal vanadates.^11,33 In contrast, X_NOX/X_NOX,0 values of Mn₁–Sb–S decreased from 1.25 to 0.7 at 5–15 hours. The resulting t_0.7 of Mn₁–Sb–S was around 1.5-fold longer and shorter than the corresponding t_0.7 values of V₂O₅–WO₃ and Mn₁–Sb–P, respectively, which demonstrated the superiority of Mn₁–Sb–P to the others in resisting AS/ABS poisons.

The overall SCR efficiencies (X_NOX) of Mn₁–Sb–P and V₂O₅–WO₃ were finally investigated via temperature-sweep SCR runs, whose environments were identical to those utilized to perform SO₂-on/off runs with SO₂ turned-on with the exception that T_REACTION values were dialled in at 150–400 °C (Fig. 9B).^11–18,33 X_NOX values maximum-achievable on Mn₁–Sb–P were 17–20% higher than those on V₂O₅–WO₃ under a SO₂-containing, wet feed gas at 180–220 °C. Interestingly, X_NOX values maximum-achievable on Mn₁–Sb–S were substantially elevated at <250 °C upon the inclusion of SO₂ in a wet feed gas (Fig. 7C and 9B) and were indistinct from those on Mn₁–Sb–P at 150–400 °C. This again could originate from the functionalization of the former's surface with additional SO_A²⁻ (BA⁻) species, whose resistance to AS/ABS, however, was lower than that of PO₄³⁻ counterparts pertaining to Mn₁–Sb–P, as verified by SO₂-on/off runs detailed above. All of these were combined to apparently prove the merit of Mn₁–Sb–P over Mn₁–Sb–S/V₂O₅–WO₃ in activating both SCR and AS/ABS pyrolysis under a SO₂-bearing, wet feed gas at low thermal energies.

3 Conclusions

To markedly improve catalytic consequences for the SCR and AS/ABS fragmentation under a wet, SO₂/H₃PO₄-containing flue gas, this study initially rationalized multiple merits expected for Mn²⁺–O–V⁵⁺ motif relative to homo-nuclear Mn²⁺–O–Mn²⁺/V⁵⁺–O–V⁵⁺ or hetero-nuclear Fe³⁺/Ni²⁺/Co²⁺/Cu²⁺–O–V⁵⁺ channels such as larger N_OL/N_OM quantities, lower E_BARRIER/hydrophilicity, or higher O_M mobility. Mn²⁺–O–V⁵⁺ motif then served to build primary building units of Mn_XV₂O_X+5 (X = 1–3), whose Mn²⁺/V⁵⁺-centered secondary building units (SBUs) are highly distinct with regard to Mn²⁺–O–V⁵⁺ connectivities. Meanwhile, Mn²⁺–O–V⁵⁺ channels are prone to dissection to generate Mn⁺, O_V, and ⁻O–V⁵⁺ (⁻O_L–V⁴⁺) species, where Mn⁺/O_V species can be modified by O₂/H₃PO₄ poison to bear mono-/bi-dentate PO₄³⁻ functionalities with terminal P⁵⁺–O²⁻ bonds acting as BA⁻ species. The resulting BA⁻-rich Mn²⁺–O–V⁵⁺ fragments subjected to PO₄³⁻ modification were combined with diverse Mn²⁺–O–V⁵⁺ connectivities stated earlier to pose bunched hierarchies on the acidic/redox traits or hydrophobicity versus X in the aspect of Mn²⁺- or V⁵⁺-centered SBUs. Interestingly, most of the aforementioned hierarchies were conflicting each other except for those on N_OL, N_OM, and E_OL values. This motivated the synthesis of PO₄³⁻-functionalized Mn_XV₂O_X+5 architectures on TiO₂ (Mn_X–P) to clarify the trends discussed above and to locate the optimum X value.

Throughout the trends on the properties of the Mn_X–P catalysts versus X, local Mn²⁺–O–V⁵⁺ connectivities of Mn²⁺-centered SBUs were as phenomenal as those of V⁵⁺-centered counterparts. This was demonstrated by the significance of the former and the latter in dictating E_NH3/N_OL and N_NH3/E_OL values of the Mn_X–P catalysts, respectively, along with their N_OM/E_OM values being mainly directed by the former. Moreover, albeit X = 1–2 were inferior to X = 3 in attaining a higher (mediated by N_NH3), the former were superior to the latter in achieving a lower E_BARRIER (mediated by E_NH3), a higher SCR activity (−r_NOX), and a higher hydrophobicity/O_M mobility (mediated by E_OM), all of which were coupled to unveil higher wet SCR performance for X = 1–2 than for X = 3. Furthermore, X = 1 outperformed X = 2 in lowering hydrophilic surface area, which corroborated that X = 1 was the optimum value. Of note was that Sb₂O₅ promoted the surface features of Mn₁–P via N_NH3 elevation and E_OM reduction, from which and hydrophobicity/O_M mobility of the resulting Mn₁–Sb–P were enhanced over the corresponding traits of Mn₁–P, respectively. Consequently, Mn₁–Sb–P exhibited higher wet SCR performance and/or higher tolerance to hydro-thermal aging than Mn_X–P, a commercial control (V₂O₅–WO₃), and a SO_A²⁻-modified analogue (Mn₁–Sb–S) we discovered previously. Of additional note was that Mn₁–Sb–P had BA⁻ species with high hydrophobicity, thereby reducing the energy needed to disintegrate pyrosulfate more pronouncedly than Mn₁–Sb–S/V₂O₅–WO₃. As a result, Mn₁–Sb–P outperformed Mn₁–Sb–S/V₂O₅–WO₃ in resisting AS/ABS poisons by hampering their deposition and/or expediting their fragmentation under a SO₂-containing wet feed gas. This study provides the guidance and methodologies concerning how to exploit poisonous H₃PO₄ as a modifier of acidic/redox sites and hydrophobicity for transition metal vanadates promising to expedite the SCR and AS/ABS fragmentation under severe environments.

Data availability

The data supporting this article have been included as part of the ESI.†

Author contributions

S. L. performed the experiments, refined/interpreted the datasets, and commented on the manuscript. S. H. P. performed the experiments. J. K. proposed the hypotheses, designed the experiments, refined/interpreted the datasets, and drafted/revised the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We are grateful to the Ministry of Science and ICT and National Research Foundation of South Korea for providing a grant for this project (#2020R1A2C2004395). We are grateful to the Basic Science Research Capacity Enhancement Project through the Korea Basic Science Institute (National research Facilities and Equipment Center) grant funded by the Ministry of Education (#2019R1A6C1010052). We are grateful to the Korea Institute for Advanced of Technology (KIAT) and the Ministry of Trade, Industry & Energy (MOTIE) of South Korea for supporting this project (#P0017363). We appreciate the Korea Institute of Science and Technology to support this project through Future R&D (#2E33171).

References

1. H. Chang, M. T. Jong, C. Wang, R. Qu, Y. Du, J. Li and J. Hao, Environ. Sci. Technol., 2013, 47, 11692–11699 CrossRef CAS PubMed .
2. K. Xie, A. Wang, J. Woo, A. Kumar, K. Kamasamudram and L. Olsson, Appl. Catal., B, 2019, 256, 117815 CrossRef CAS .
3. W. Chen, R. Zou and X. Wang, ACS Catal., 2022, 12, 14347–14375 CrossRef CAS .
4. Z. Liu, Y. Fu, Y. Shan, T. Zhang, Y. Wang, J. Du, Y. Sun, G. He, X. Shi, Y. Yu and H. He, ACS Catal., 2024, 14, 1678–1689 CrossRef CAS .
5. S. Sethuram Markandaraj, T. Muthusamy and S. Shanmugam, Adv. Sci., 2022, 9, 2201410 CrossRef CAS PubMed .
6. Y.-i. Kwon, S. K. Kim, Y. B. Kim, S. J. Son, G. D. Nam, H. J. Park, W.-C. Cho, H. C. Yoon and J. H. Joo, ACS Energy Lett., 2021, 6, 4165–4172 CrossRef CAS .
7. J. A. Dean, Lange's Handbook of Chemistry, McGraw-Hill Professional, New-York, 15th edn, 1998 Search PubMed .
8. M. Jabłońska, M. E. Potter and A. M. Beale, ChemCatChem, 2024, 16, e202301214 CrossRef .
9. M. Li, J. Henao, Y. Yeom, E. Weitz and W. M. H. Sachtler, Catal. Lett., 2004, 98, 5–9 CrossRef CAS .
10. G. Zhou, B. Zhong, W. Wang, X. Guan, B. Huang, D. Ye and H. Wu, Catal. Today, 2011, 175, 157–163 CrossRef CAS .
11. J. Kim, D. H. Kim and H. P. Ha, J. Hazard. Mater., 2020, 397, 122671 CrossRef CAS PubMed .
12. J. Kim, K. B. Nam and H. P. Ha, J. Hazard. Mater., 2021, 416, 125780 CrossRef CAS PubMed .
13. J. Kim, S. Lee and H. P. Ha, ACS Catal., 2021, 11, 767–786 CrossRef CAS .
14. J. Kim, D. H. Kim, J. Park, K. Jeong and H. P. Ha, ACS Catal., 2022, 12, 2086–2107 CrossRef CAS .
15. S. Lee, J.-H. Lee, H. P. Ha and J. Kim, Chem. Mater., 2022, 34, 1078–1097 CrossRef CAS .
16. S. Lee, H. P. Ha, J.-H. Lee and J. Kim, J. Mater. Chem. A, 2023, 11, 12062–12079 RSC .
17. S. Lee, H. P. Ha, J.-H. Lee and J. Kim, J. Hazard. Mater., 2023, 460, 132278 CrossRef CAS PubMed .
18. S. Lee, J. Choi, H. P. Ha, J.-H. Lee, J. Park and J. Kim, ACS Catal., 2024, 14, 3349–3368 CrossRef CAS .
19. S. Xiong, Y. Liao, X. Xiao, H. Dang and S. Yang, Catal. Sci. Technol., 2015, 5, 2132–2140 RSC .
20. S. Yang, S. Xiong, Y. Liao, X. Xiao, F. Qi, Y. Peng, Y. Fu, W. Shan and J. Li, Environ. Sci. Technol., 2014, 48, 10354–10362 CrossRef CAS PubMed .
21. L. Kang, L. Han, P. Wang, C. Feng, J. Zhang, T. Yan, J. Deng, L. Shi and D. Zhang, Environ. Sci. Technol., 2020, 54, 14066–14075 CrossRef CAS PubMed .
22. T. H. Vuong, J. Radnik, J. Rabeah, U. Bentrup, M. Schneider, H. Atia, U. Armbruster, W. Grünert and A. Brückner, ACS Catal., 2017, 7, 1693–1705 CrossRef CAS .
23. J. Mosrati, A. M. Abdel-Mageed, T. H. Vuong, R. Grauke, S. Bartling, N. Rockstroh, H. Atia, U. Armbruster, S. Wohlrab, J. Rabeah and A. Brückner, ACS Catal., 2021, 11, 10933–10949 CrossRef CAS .
24. X. Lei, C. Jiang, Q. Han, X. Zhang, K. Zhao, N. Yan, H. Guo, B. Tang, Y. Li, Y. Cui, X.-Z. Fu, J. Li and Y. Sun, ACS Catal., 2024, 4523–4535,  DOI:10.1021/acscatal.3c05789 .
25. X. Yin, H. Han and A. Miyamoto, Phys. Chem. Chem. Phys., 2000, 2, 4243–4248 RSC .
26. A. Miyamoto, Y. Yamazaki, T. Hattori, M. Inomata and Y. Murakami, J. Catal., 1982, 74, 144–155 CrossRef CAS .
27. Y. Liu, C. Wen, Y. Guo, G. Lu and Y. Wang, J. Phys. Chem. C, 2010, 114, 9889–9897 CrossRef CAS .
28. C. Li, J. He, Y. Xiao, Y. Li and J.-J. Delaunay, Energy Environ. Sci., 2020, 13, 3269–3306 RSC .
29. C. Tantardini and A. R. Oganov, Nat. Commun., 2021, 12, 2087 CrossRef CAS PubMed .
30. B. Zoellner, E. Gordon and P. A. Maggard, Dalton Trans., 2017, 46, 10657–10664 RSC .
31. C. Zheng, B. Jian, J. Zhong and S. Huang, J. Alloys Compd., 2023, 934, 168018 CrossRef CAS .
32. O. Clemens, J. Rohrer and G. Nénert, Dalton Trans., 2016, 45, 156–171 RSC .
33. H. J. An, D. H. Kim, H. P. Ha and J. Kim, J. Mater. Chem. A, 2021, 9, 8350–8371 RSC .
34. Q. Tang, K. Huang, G. Qian and B. C. Benicewicz, J. Power Sources, 2013, 229, 36–41 CrossRef CAS .
35. S. Palaniappan, C. Saravanan, C. A. Amarnath and V. J. Rao, Catal. Lett., 2004, 97, 77–81 CrossRef CAS .
36. Y. J. Choe, S. H. Kim, K. Jeong and J. Kim, Chem. Eng. J., 2023, 455, 140537 CrossRef CAS .
37. K. G. Gorbovskiy, A. I. Kazakov, A. M. Norov and A. I. Mikhaylichenko, Russ. J. Appl. Chem., 2018, 91, 1057–1067 CrossRef CAS .
38. M. Statheropoulos and S. A. Kyriakou, Anal. Chim. Acta, 2000, 409, 203–214 CrossRef CAS .
39. A. Marcilla, M. I. Beltran, A. Gómez-Siurana, I. Martinez-Castellanos, D. Berenguer, V. Pastor and A. N. García, J. Anal. Appl. Pyrolysis, 2015, 112, 48–55 CrossRef CAS .
40. S. Giraud, S. Bourbigot, M. Rochery, I. Vroman, L. Tighzert, R. Delobel and F. Poutch, Polym. Degrad. Stab., 2005, 88, 106–113 CrossRef CAS .
41. J. Kim, D. W. Kwon, S. Lee and H. P. Ha, Appl. Catal., B, 2018, 236, 314–325 CrossRef CAS .
42. J. Kim, D. H. Kim, D. W. Kwon and H. P. Ha, Catal. Sci. Technol., 2018, 8, 4774–4787 RSC .
43. J. Kim, S. Lee, D. W. Kwon, K.-Y. Lee and H. P. Ha, Appl. Catal., A, 2019, 570, 355–366 CrossRef CAS .
44. K. Yu, Y. Liu, D. Lei, Y. Jiang, Y. Wang, Y. Feng, L.-L. Lou, S. Liu and W. Zhou, Catal. Sci. Technol., 2018, 8, 2299–2303 RSC .
45. B. Zhao, R. Ran, X. Guo, L. Cao, T. Xu, Z. Chen, X. Wu, Z. Si and D. Weng, Appl. Catal., A, 2017, 545, 64–71 CrossRef CAS .
46. Q. Wan, Q. Yao, L. Duan, X. Li, L. Zhang and J. Hao, Environ. Sci. Technol., 2018, 52, 2981–2987 CrossRef CAS PubMed .
47. B. von Boehn, T. O. Menteş, A. Locatelli, A. Sala and R. Imbihl, J. Phys. Chem. C, 2017, 121, 19774–19785 CrossRef CAS .
48. J. S. Geelhoed, T. Hiemstra and W. H. Van Riemsdijk, Environ. Sci. Technol., 1998, 32, 2119–2123 CrossRef CAS .
49. M. I. Tejedor-Tejedor and M. A. Anderson, Langmuir, 1990, 6, 602–611 CrossRef CAS .
50. W. Li, X. Feng, Y. Yan, D. L. Sparks and B. L. Phillips, Environ. Sci. Technol., 2013, 47, 8308–8315 CAS .
51. H. Puspito Buwono, Y. Omori, N. Shioya, H. Yoshida, S. Hinokuma, Y. Nagao, K. Iwashina, Y. Endo, Y. Nakahara and M. Machida, Catal. Sci. Technol., 2019, 9, 5447–5455 RSC .
52. Y. Weng, J. Vekeman, H. Zhang, L. Chou, M. Elskens and F. Tielens, Chemosphere, 2020, 261, 127776 CrossRef CAS PubMed .
53. X. Zhu, Y. Liu, F. Qian, H. Shang, X. Wei, S. Zhang, J. Chen and Z. J. Ren, J. Mater. Chem. A, 2018, 6, 3705–3713 RSC .
54. W. Huang, M. Schopfer, C. Zhang, R. C. Howell, L. Todaro, B. A. Gee, L. C. Francesconi and T. Polenova, J. Am. Chem. Soc., 2008, 130, 481–490 CrossRef CAS PubMed .
55. W. Zhao, S. Dou, K. Zhang, L. Wu, Q. Wang, D. Shang and Q. Zhong, Chem. Eng. J., 2019, 364, 401–409 CrossRef CAS .
56. J. Wang, Z. Yan, L. Liu, Y. Chen, Z. Zhang and X. Wang, Appl. Surf. Sci., 2014, 313, 660–669 CrossRef CAS .
57. Y. Xin, H. Li, N. Zhang, Q. Li, Z. Zhang, X. Cao, P. Hu, L. Zheng and J. A. Anderson, ACS Catal., 2018, 8, 4937–4949 CrossRef CAS .
58. F. Liu, K. Asakura, H. He, W. Shan, X. Shi and C. Zhang, Appl. Catal., B, 2011, 103, 369–377 CrossRef CAS .
59. H. Hu, S. Cai, H. Li, L. Huang, L. Shi and D. Zhang, ACS Catal., 2015, 5, 6069–6077 CrossRef CAS .
60. Z. Liu, S. Zhang, J. Li and L. Ma, Appl. Catal., B, 2014, 144, 90–95 CrossRef CAS .
61. F. Babou, G. Coudurier and J. C. Vedrine, J. Catal., 1995, 152, 341–349 CrossRef CAS .
62. X. Yang, K. Liu, X. Han, J. Xu, M. Bian, D. Zheng, H. Xie, Y. Zhang and X. Yang, J. Hazard. Mater., 2023, 459, 132209 CrossRef CAS PubMed .
63. M. Klimczak, P. Kern, T. Heinzelmann, M. Lucas and P. Claus, Appl. Catal., B, 2010, 95, 39–47 CrossRef CAS .
64. O. Kröcher and M. Elsener, Appl. Catal., B, 2008, 77, 215–227 CrossRef .
65. X. Xu, X. Han, Y. Zheng, W. Zhou, K. Davey and S.-Z. Qiao, Chem Catal., 2023, 3, 100794 CrossRef CAS .
66. W. Yang, J. Gong, X. Wang, Z. Bao, Y. Guo and Z. Wu, ACS Catal., 2021, 11, 12446–12468 CrossRef CAS .
67. P. Auvinen, J. T. Hirvi, N. M. Kinnunen and M. Suvanto, ACS Catal., 2020, 10, 12943–12953 CrossRef CAS .
68. B. Wang, M. Wang, L. Han, Y. Hou, W. Bao, C. Zhang, G. Feng, L. Chang, Z. Huang and J. Wang, ACS Catal., 2020, 10, 9034–9045 CrossRef CAS .
69. X. Liu, P. Wang, Y. Shen, S. Bi, W. Ren and D. Zhang, ACS Catal., 2022, 12, 11306–11317 CrossRef CAS .

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section; structural features and properties versus X for MnXV2OX+5 architectures; properties, EDX mapping images, XRD patterns, XP/31P MAS NMR spectra, NH3-TPD/O2-TPD/SO2-TPD/H2-TPR/TGA-MASS profiles, and H2O isotherms for the catalysts; −rNOX/−rABS values and Arrhenius plot for the catalysts. See DOI: https://doi.org/10.1039/d4ta03928a

This journal is © The Royal Society of Chemistry 2024