The [(NH₃)₄Cu₂O₂]²⁺-Peroxo Complex as the Key Intermediate for NH₃-SCR Activity and Deactivation of Cu-CHA Catalysts.

Graphical Abstract

The chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex determines the NH₃-SCR reaction over Cu-CHA zeolite-based catalysts at low temperature (150–300 °C). The reactivity of the complex is analyzed in relation to SCR activity (reaction with NO), inhibition (reaction with NH₃) and poisoning (reaction with SO₂).

Abstract

In NH₃-SCR over Cu-CHA catalysts in the low-temperature range 150–300 °C, the activation of oxygen occurs via oxidation of a pair of mobile (NH₃)₂Cu^I-complexes located in the cages of the zeolite. In this step, a reactive [(NH₃)₄Cu₂O₂]²⁺-peroxo complex (μ-η²,η²-peroxo diamino dicopper(II)-complex) is formed. The chemistry of this complex determines several catalytic properties of the Cu-CHA catalyst. The reaction of NO with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex governs the NH₃-SCR activity. A reaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with ammonia hinders the reaction of NO with the complex, thus leading to an inhibition of the NH₃-SCR reaction. Finally, the deactivation in presence of SO₂ is due to a reaction of SO₂ with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, leading to the formation of Cu−S compounds in the catalyst. In this review, the characterization and the reactions of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with NO, NH₃, and SO₂, and mean-field kinetic models based on first principles calculations for NH₃-SCR activity and SO₂ poisoning are discussed.

1 Introduction

Since the discovery of Cu-chabazite materials (Cu-CHA, also referred to as Cu-SSZ-13) as effective catalysts for selective catalytic reduction by ammonia (NH₃-SCR),^1-3 these have become a widely applied catalyst for the abatement of NOx in diesel-powered passenger cars and heavy-duty vehicles. In the NH₃-SCR reaction, NOx is reduced to nitrogen and water by ammonia, which is added to the exhaust gas via injection of a urea solution (AdBlue^® or Diesel Exhaust Fluid (DEF)). Cu-CHA catalysts feature excellent NH₃-SCR activity at temperatures in the 150–300 °C range, and a good thermal stability. This means that they generally can be operated in the temperature window 150–550 °C, as is required for application in exhaust systems. The thermal stability of Cu-CHA catalysts ensures that the Cu-CHA catalysts can be exposed to temperatures up to about 800 °C, without significant degradation of the zeolite material.^{4, 5} This makes these catalysts very robust in the harsh and dynamic environment in exhaust systems. Another attractive property of Cu-CHA catalysts for NH₃-SCR is that such catalysts have a low selectivity for N₂O formation, as compared to other Cu-zeolite systems.^{5, 6} Because N₂O is a strong greenhouse gas with a global warming potential of about 300,⁷ the use of Cu-CHA catalysts also contributes to minimize greenhouse gas emissions from exhaust systems.

In the NH₃-SCR reaction, the conversion of NOx – which in this article refers to the NO and NO₂ compounds combined – to N₂ and H₂O by NH₃ takes place according to the standard-SCR reaction, which follows the equation: (Eq 1:

$$ (1)

When NO₂ is present, the reaction is accelerated, and proceeds according to the fast-SCR reaction:

$$ (2)

In situations where the concentration of NO₂ exceeds the NO concentration, the NO₂-SCR reactions also contribute to the conversion of NOx:⁸

$$ (3)

$$ (4)

$$ (5)

As illustrated by the equations above, the NO₂-SCR reactions, taking place in the absence of NO, describe a more complex reaction scheme. Because the exhaust gas of fuel-lean combustion engines, such as diesel engines, typically contains 5–10 % oxygen, together with a NO₂/NO ratio lower than 1, the standard-SCR (Eq. 1) and fast-SCR (Eq. 2) reactions are the most relevant reactions for the abatement of NOx in diesel exhaust systems. Here, we focus exclusively on the standard-SCR reaction, and, in particular, the mechanism of that reaction in the 150–300 °C temperature range.

As the reaction products in the standard-SCR reaction do not contain an O−O bond (Eq. 1), the NH₃-SCR reaction implies a dissociation of the O−O bond in the O₂ molecule. In Cu-CHA catalysts, the dissociation of the O₂ molecule involves the formation of a μ-η²,η²-peroxo diamino dicopper(II) ([(NH₃)₄Cu₂O₂]²⁺-peroxo) complex. This [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is key for the NH₃-SCR activity over Cu-CHA catalysts,^{9, 10} but, at the same time, it is also an important channel for deactivation by SO₂. The NH₃-SCR activity is related to the reactivity of this complex towards NO and NH₃; the dissociation of the O₂ molecule then takes place in the reactions of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with these molecules. The deactivation of Cu-CHA catalysts by SO₂ is a consequence of the reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with SO₂.^11-13 These aspects illustrate how the chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is pivotal for the fundamental understanding of Cu-CHA catalysts for NH₃-SCR. In this review, we discuss the identification of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex as a reaction intermediate in the standard-SCR reaction (Eq. 1), and how its chemistry determines the kinetics of that reaction, the formation of N₂O, and deactivation by SO₂.

2 Importance of Mobility and Cu-Pair Formation for Activation of O₂

In the NH₃-SCR reaction cycle, the Cu ions in the zeolite undergo alternating oxidation and reduction steps, whereby the oxidation state changes between Cu^I and Cu^II. In fact, the oxidation and reduction steps have been executed separately.^14-21 For the oxidation half-cycle, the catalyst is exposed to O₂ or a mixture of O₂ and NOx; the reduction takes place in a mixture of NO and NH₃ in the absence of O₂. Because O₂ does not react with any of the Cu^II species in the NH₃-SCR reaction cycle, the activation of O₂ implies an oxidation of Cu^I.^{9, 14, 15, 17, 18, 22} Moreover, this is the only possible reaction step to oxidize the Cu^I to Cu^II in the standard-SCR reaction cycle, as neither NO nor NH₃ undergo a reaction with Cu^I. This strict selectivity of the reactivity of O₂ to Cu^I is the reason why the oxidation and reduction half-cycles in the NH₃-SCR reaction can be done separately.^{14-16, 23-27}

The reaction of an O₂ molecule with Cu^I is not a straightforward elementary step, as follows from the stoichiometry of the NH₃-SCR reaction. According to Eq. 1, a single O₂ molecule reacts with four molecules of NO and NH₃, to form H₂O and nitrogen. That process requires the transfer of four electrons to the oxygen from the other reactants. Since a single Cu^I ion can only transfer one electron, it follows that the reaction of O₂ requires either a reaction of a single O₂ molecule with four Cu ions in the zeolite, or the participation of an alternative electron donor. The formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex suggests, that the actual NH₃-SCR reaction cycle involves a reaction of an O₂ molecule with two Cu ions, accounting for the transfer of two electrons. The remaining two electrons are eventually delivered by the NO reactant. An early indication that two Cu ions are required in the NH₃-SCR reaction follows from the observation that the catalytic activity of Cu-CHA catalysts show a 2^nd order dependence on the Cu content.^{26, 28-32}

Density functional theory (DFT) calculations indicate, that the activation of an O₂ molecule becomes significantly easier when it can interact with two Cu ions simultaneously, resulting in formation of binuclear or multinuclear Cu_xO_y species.^{33, 34} Reactions with O₂ and two Cu^I ions bound to the zeolite framework are possible, but generally they require suitable configurations of the Cu ions, dictated by the crystal structure and Si/Al ratio of the zeolite.³⁵ This means that stabilization of Cu_xO_y species requires certain framework structural motifs. As an example, a mono(μ-oxo)dicopper core in ZSM-5 should be located at the intersection of straight and sinusoidal ten-membered rings, formed by O₂ (or N₂O) reaction with two Cu^I ions at distance of 4.17 Å.^{36, 37} A similar structure can be formed in Cu-mordenite at a different structural motif, i. e. at the intersection of the side pocket with the 12 and 8 member rings channels, with an adjacent Al framework atoms distance around 7.55 Å.³⁸ In Cu-CHA, O₂ can be activated by two Cu^I ions, when both ions are located in an 8-membered ring.³⁹

With Cu-ions fixed to the zeolite crystal, and a random distribution of the possible Cu-exchange sites generated by the Al atoms in the zeolite framework, the number of possible sites with two Cu-ions that can “fit” a single O₂ molecule is limited. If the Cu ions become mobile, it is easier to form suitable Cu-pairs. Some mobility of Cu ions has been observed during thermal treatment of the catalyst,^40-43 where Cu^I ions formed by ‘self-reduction’ of Cu^II during heating in inert gas migrate from an 8-membered ring to the double-6-rings typical of the CHA structure.⁴² Similarly, a redistribution of Cu^II ions initially present as Z-Cu^IIOH in 8-membered rings (Z represents a negative charge on the zeolite framework) to Z₂-Cu^II in the 6-membered rings has been observed during thermal treatment in air.⁴³

In Cu-CHA based catalysts, the Cu ions are coordinated by NH₃, NO₃⁻, NO₂⁻, and OH⁻ ligands under conditions for NH₃-SCR below 300 °C, forming Cu-complexes.^{14, 15, 17, 18, 44} In particular, the Cu^I species is a linear (NH₃)₂Cu^I complex, which can be identified by a characteristic pre-edge peak at the Cu−K edge in the X-ray absorption (XAS) spectrum.^{15, 17, 45, 46} The absence of a clear Cu−Si or Cu−Al coordination in the Extended X-ray Absorption Fine Structure (EXAFS) after the formation of this complex indicates a weak interaction of the (NH₃)₂Cu^I complex with the zeolite framework.^{15, 17, 47} Ab-initio molecular dynamics calculations show that this weak interaction allows for some mobility of the complex in the zeolite cages.^{17, 47-49} Furthermore, gas phase NH₃ induces a solid-state ion exchange process in a physical mixture of CuO or Cu₂O and a H-CHA zeolite at temperatures as low as 250 °C, under the formation of the (NH₃)₂Cu^I complex.^{48, 50-52} This observation provides experimental evidence for an enhanced mobility of the (NH₃)₂Cu^I complex under SCR conditions.

In the NH₃-SCR reaction over Cu-CHA catalysts via Cu-pair formation, the mobility of the (NH₃)₂Cu^I complexes facilitates the activation of O₂.^{18, 30, 34, 47} As a consequence, the oxidation step of the linear (NH₃)₂Cu^I complexes by O₂ shows a 2^nd order dependence on the Cu-content of the catalyst.^{16, 30, 32} When the Cu content in the catalyst decreases, an increasing fraction of the (NH₃)₂Cu^I complexes does not react with O₂, and remains in the reduced state.^{30, 53} Calculations of the molecular dynamics of the system indicate that the (NH₃)₂Cu^I complexes can move up to about 9 Å away from their anchor point in the zeolite.^{17, 30} At lower Cu loadings, the average distance between two Cu ions becomes larger, and therefore the propensity of suitable (NH₃)₂Cu^I pairs for O₂ activation decreases in catalysts with low Cu content, resulting in the observed 2^nd order behavior. In transient kinetic models, this 2^nd order behavior has been explicitly implemented, to produce accurate descriptions of the individual oxidation and reduction steps in the NH₃-SCR cycle.^{16, 20, 22, 26, 27, 32}

Direct experimental evidence for importance of the mobility of (NH₃)₂Cu^I complexes for the oxidation step in the NH₃-SCR reaction cycle is provided by a calorimetry study of the reactivity of framework-bound Cu^I and (NH₃)₂Cu^I, showing a stronger reaction of oxygen with the (NH₃)₂Cu^I complexes.⁵⁴ Figure 1 shows that, at temperatures below 200 °C, the heat release upon reaction of oxygen with (NH₃)₂Cu^I complexes (2 to 3 J/g_cat) is clearly higher compared to framework Cu^I. These results indicate that the presence of (NH₃)₂Cu^I complexes enhances the reaction with O₂, in line with the idea that the mobile (NH₃)₂Cu^I complexes facilitate the Cu-pair formation that is needed for the reaction with oxygen. The increase in heat release with the reaction temperature indicate that the reactions of oxygen with the (NH₃)₂Cu^I complexes and framework Cu^I species are activated processes, with estimated activation energies of 15 kJ/mol and 39 kJ/mol,⁵⁴ respectively. The decrease in heat release for the (NH₃)₂Cu^I complexes above 200 °C correlates well with the temperature range where the decomposition of the (NH₃)₂Cu^I complexes takes place,^55-57 leading to a slower oxidation reaction, as the Cu loses its mobility. The oxidation of the framework-bound Cu^I species becomes more prominent above 200 °C, suggesting that the oxidation of framework-bound Cu^I potentially plays a role in the NH₃-SCR reaction at higher temperatures.^{18, 39, 58} Such a change in reaction mechanism explains the often observed decrease in NOx conversion between 300 and 400 °C (“seagull” effect).³⁹

Figure 1

Heat release upon exposure to 500 ppm O₂ at various temperatures between 75 and 275 °C over the (NH₃)₂Cu^I complexes (red dots/solid line) and the framework-bound Cu^I species (blue dots/dashed line) in 3 wt% Cu-CHA catalyst. Reprinted with permission from Ref. [54]. Copyright © 2021 X. Wang et al. ChemCatChem published by Wiley-VCH GmbH.

Finally, the results shown in Figure 1 provide an efficient experimental procedure to prepare the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex in a Cu-CHA catalyst. First, the (NH₃)₂Cu^I complex is formed by reduction of the catalyst at 200–250 °C, to ensure complete reduction. In a second step, the catalyst is exposed to at least 4 % O₂ at 200 °C,⁵⁴ in order to form the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. This procedure is of great value in experimental studies of the chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex.

3 Identification of the [(NH₃)₄Cu₂O₂]²⁺-Peroxo Complex

The most probable products of a reaction of O₂ with a pair of (NH₃)₂Cu^I complexes ions are (1) a trans-μ-1,2-peroxo diamino dicopper(II) complex, with an end-on orientation of the O₂ molecule between the Cu-ions, (b) a μ-η²,η²-peroxo diamino dicopper(II) complex (the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex), with a side-on orientation of the O₂ molecule, without a complete dissociation of the O₂ bond, or (c) bis-μ-oxo diamino dicopper(III)- complex, a side-on orientation of the O₂ molecule, combined with a O−O bond dissociation. These structures are shown in Figure 2. Based on DFT calculations, EXAFS and UV-Vis spectroscopy, it has been shown that the μ-η²,η²-peroxo diamino dicopper(II) complex is formed in the reaction of the linear (NH₃)₂Cu^I complexes with O₂.^{24, 59, 60}

Figure 2

Ball-and-stick drawings of the three most probable [Cu₂(NH₃)₄O₂]²⁺ complexes: (a) trans-μ-1,2-peroxo diamino dicopper(II) (end-on), (b) μ-η²,η²-peroxo diamino dicopper(II) (side-on), and c) bis-μ-oxo diamino dicopper(III). Atom color code: Cu, green; H, white; O, red; N, blue. Reproduced from Ref. [24]. Copyright © 2020 American Chemical Society. Publication licensed under CC-BY 4.0.

The density functional theory (DFT) has become the standard approach to calculate reaction energies in heterogeneous catalysis, as it generally combines computational efficiency with reasonable accuracy.⁶¹ However, the problem to determine the structure of the reaction product formed in the oxidation of a pair of (NH₃)₂Cu^I complexes with DFT is that the result critically depends on the choice of the exchange-correlation functional. This is illustrated in Figure 3, which shows calculated reaction paths for the reaction of a pair of (NH₃)₂Cu^I complexes with O₂, for different choices of the functional.⁵⁹ The oxygen dissociation path is divided in steps with five structures denoted C1 to C5. In general, all functionals indicate that Cu-pairs are favorable for oxygen adsorption. Some of the functionals based on the Perdew-Burke-Ernzerhof (PBE) formula⁶² predict the formation of a bis-μ-oxo diamino dicopper(III) complex (Figure 3 C5/Figure 2c). In contrast, the Bayesian error estimation functional (BEEF), PBE functionals augmented with a Hubbard-U term (PBE+U), and PBE+U functionals with corrections to account for Van der Waals interactions (PBE+U+D2) and (PBE+U+D3), and finally a hybrid functional HSE06, all point to the formation of a μ-η²,η²-peroxo diamino dicopper(II) complex (Figure 3 C4/Figure 2b). Based on calculations of similar Cu,O-complex structures in enzymes with experimentally verified structures, it could be shown that the PBE+U+D3 approach is a reasonable compromise between computational efficiency and accuracy.⁵⁹ Using this functional, with U=6 eV for the Hubbard term,⁵⁹ we find that the oxidation of a pair of (NH₃)₂Cu^I complexes with oxygen results in the side-on configuration, forming a μ-η²,η²-peroxo diamino dicopper(II) complex (Figure 3 C4/Figure 2b).

Figure 3

Potential energy landscapes for oxidation of (NH₃)₂Cu^I-pairs in CHA calculated by different functionals. The structures considered in each step are shown in the top. TS stands for the transition state from C2 to C3, which include a triplet to singlet transition. The barriers are listed below the curves with the corresponding colors. The stars in the potential energy landscape represent the cross-over between triplet and singlet state for O₂ adsorbed on the pair. All energies are zero-point energy corrected. Atom color codes: copper (pink), aluminum (purple), nitrogen (blue), silicon (yellow), oxygen (red) and hydrogen (white). Reproduced from Ref. [59], with permission from the PCCP Owner Societies.

The fact that the results depend on the choice of functional illustrates that DFT methods based on the Generalized Gradient Approximation (GGA) are not sufficient to describe the system. The main reason is the poor description of charge transfer processes due to the self-interaction error, thus incomplete cancellation of the electron self-interaction between the direct Coulomb interaction and the exchange correlation functional. Moreover, GGA functionals do not account for the van der Waals interactions. Because the [Cu₂O₂(NH₃)₄]²⁺ peroxo complex in CHA is floating in the van der Waals interactions and Coulomb interactions from the zeolite framework, a functional accounting for these interaction is required. The van der Waals interactions can be accounted for by semi-empirical methods, corresponding to the D2 and D3 functionals, or through a proper nonlocal correlation term in the exchange-correlation functional (vdW-DF),^63-66 which was used in the results denoted CX,⁶⁷ or the BEEF functional.⁶⁸

The preference for the side-on or bis-configuration is largely determined by the way the degree of electron delocalization is calculated. The functionals with large electron delocalization yield the bis-μ-oxo diamino dicopper(III), while the functionals where the delocalization is suppressed favor the μ-η²,η²-peroxo diamino dicopper(II) side-on configuration. The Hubbard-U term controls the electron localization for the PBE functionals, and in the HSE06 hybrid functional, the electron localization is favored by including a part of Fock-exchange. Apparently, the PBE+U−D3 functional results in an adequate description of the Van der Waals interactions and electron delocalization for the Cu−O interactions in Cu-CHA materials.

Whether a bis-μ-oxo diamino dicopper(III) or a μ-η²,η²-peroxo diamino dicopper(II) complex is formed upon oxidation of a pair of Cu^I depends on the ligands that are attached to the Cu ions.^{59, 69} In principle, the two complexes form an equilibrium, which can be shifted by changing the ligands and environment of the Cu−O cores.⁶⁹ In fact, a transition state energy of 0.16 eV was calculated for this transition in a Cu-CHA, using a - less appropriate - BEEF-vdW functional.³⁴ With the NH₃-ligands, the μ-η²,η²-peroxo diamino dicopper(II) is clearly favored in Cu-CHA catalysts, which is further confirmed by spectroscopy, as shown below.

Experimental evidence for the formation of a μ-η²,η2-peroxo diamino dicopper(II) complex, as derived from DFT, comes primarily from in-situ X-ray absorption spectroscopy (XAS) and DR-UV-vis spectroscopy.²⁴ The X-Ray Near Edge spectra (XANES) of a reduced Cu-CHA catalyst (Si/Al=15; Cu/Al=0.5, corresponding to a Cu density of ~0.4 Cu/1000 ų) shows that most (84±8 %) of the Cu^I is oxidized to Cu^II when exposed to O₂ at 200 °C, in good agreement with the mechanism based oxygen activation on mobile (NH₃)₂Cu^I complexes.³⁰ The Cu^II species formed upon exposure to O₂ at 200 °C shows distinct XANES, EXAFS and DR-UV-Vis spectroscopic features (red curves in Figure 4a-c), that are different from those observed after pretreatment in O₂ at 400 °C (grey dashed curves in Figure 4a-c). Because the latter are due to monomeric and multimeric Cu^II species coordinated to the zeolite framework,^{70, 71} this indicates that the oxidation with O₂ at 200 °C of (NH₃)₂Cu^I complexes leads to a different Cu species.

Figure 4

a) XANES, (b) EXAFS and (c) UV-Vis spectra of Cu-CHA (Si(Al=15, Cu/Al=0.5) after pretreatment in O₂ (grey dashed curve), reduction in NO/NH₃ (blue) and subsequent reaction with O₂ (red) at 200 °C. The top panel reports a pictorial representation of the dominant Cu-species expected at each step, using the same color code. Atom color code in the structures: Cu, green; H, white; O, red; N, blue, Al, yellow; Si, grey. (d–f) Moduli of EXAFS WTs in the 2–4 Å R-space range for (d) pre-treatment in O₂, (e) reduction in NO/NH₃, (f) subsequent reaction with O₂; part (f) also includes the corresponding EXAFS spectra in k- and R-space (conventional FT), as well as the backscattering amplitude function F(k) of relevant atoms. Summary of results presented in Ref. [24].

A Wavelet Transform (WT) analysis of the EXAFS data yields a first indication for the formation of a Cu-complex with two Cu centers. This analysis is based on a 2D representation of the EXAFS data, showing the dependence in both R-space and k-space. In particular, the k-space dependency allows for a discrimination of the scattering contributions involving a Cu-atom from those that exclusively involve the lighter elements, such as O/N or Al/Si, due to element-specific k-space dependency of the EXAFS back-scattering amplitude functions F(k) (Figure 4f, top panel). In particular, a sub-lobe centered around k=7 Å⁻¹ is diagnostic for a Cu−Cu scattering path. Figure 4e and f show a clear increase in this Cu−Cu scattering upon oxidation of the linear (NH₃)₂Cu^I complexes by oxygen at 200 °C, indicating the formation of a complex with two Cu centers. The calculated EXAFS scattering for the three structure models shown in Figure 2 indicate that the measured data are most consistent with the expected EXAFS for the μ-η²,η²-peroxo diamino dicopper (II) complex, with a Cu−Cu distance of ~3.40 Å.

In DR-UV-Vis spectra, the formation of the μ-η²,η²-peroxo diamino dicopper (II) complex is indicated by a broad band in the d-d region centered at 13850 cm⁻¹ (~720 nm) and a red-shift of the LMCT transitions (ligand-to-metal charge transfer), that differs from that associated with the framework-coordinated Cu^II species in the pre-treated Cu-CHA catalyst (Figure 4c).²⁴ Upon exposure of Cu-CHA and Cu-AEI catalysts to an NH₃-SCR feed at 250 °C, a distinct band in the 370–350 nm range of the DR-UV-Vis spectrum becomes most prominent. With the aid of DFT calculations and comparison to bio-inorganic cores,⁷² this band can be assigned to the μ-η²,η²-peroxo diamino dicopper (II) complex,⁶⁰ further corroborating the role of this species as a key intermediate in the SCR cycle.

4 Structure of the[(NH₃)₄Cu₂O₂]²⁺-Peroxo Complex

Because the [Cu₂O₂(NH₃)₄]²⁺ peroxo complex is formed in a reaction between mobile (NH₃)₂Cu^I complexes and O₂, and the mobility of the (NH₃)₂Cu^I complexes is affected by the density and distribution of the Al atoms in the zeolite framework, it is reasonable to assume that the formation and structure of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex can be affected by the composition and local environment in the zeolite. Indeed, results from in-situ XAS, coupled with WT-EXAFS and Machine Learning (ML) assisted fitting methods show that the structure of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex and NH₃-SCR performance are affected by the Si/Al ratio and Al distribution in the zeolite.⁷³

The Cu−K-edge XANES spectra for the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex in for three Cu-CHA catalysts (1.51 wt% Cu, Si/Al=5, 2.64 wt% Cu, Si/Al=15, and 1.71 wt% Cu, Si/Al=29) are shown in Figure 5a. We note that after the formation of the the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex at 200 °C, the fraction of Cu^I (pie charts Figure 5a) for Si/Al=29 is higher than that for Si/Al=5, even though the Cu densities in these catalysts are comparable. The data by Paolucci et al.³⁰ follow the same trend. The turn over frequency increases with decreasing Cu^I fraction indicating that a more efficient oxidation of (NH₃)₂Cu^I by O₂ activation is beneficial for the low temperature NH₃-SCR performance.⁷³

Figure 5

(a) In situ Cu K-edge XANES for Cu-CHA catalysts after pre-treatment, reduction, and oxidation: (a) 1.51 wt% Cu, Cu/Al=0.11, Si/Al=5 (“0.1_5”), (b) 2.64 wt% Cu, Cu/Al=0.47, Si/Al=15 (“0.5_15”) and (c) 1.71 wt% Cu, Cu/Al=0.56, Si/Al=29 (“0.6_29”). Pie charts illustrate Cu^I/Cu^II percentages evaluated by XANES Linear Combination Fit (LCF) at the end of the oxidation step. (b). EXAFS-WT maps after reduction (left) and oxidation (right), magnified in the 2−4 Å range. (c) Example of ML-assisted EXAFS fitting for 1.51 wt% Cu / Si/Al=5 with molecular models of the employed structural components and their percentages determined for the three catalysts after the oxidation step, illustrated by the pie charts. Summary of results presented in Ref. [73].

A more detailed structural analysis based on the WT-EXAFS maps obtained after oxidation in O₂ (Figure 5b) reveals the presence of dicopper(II) complexes in all catalysts. Additionally, two distinct Cu−Cu interatomic distances can be identified (LR and HR labels with white arrows in Figure 5b), indicating a bimodal distribution of the Cu−Cu distance along the R direction, which suggests the presence of two different structures of the dicopper(II) complexes. A novel Machine Learning-assisted EXAFS analysis, accounting for the presence of (NH₃)₂Cu^I and a bimodal distribution of Cu−Cu distances in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, confirmed the presence of two different configurations corresponding to a ′bent′ and a ′planar′ configuration.⁷³ At Si/Al=5, the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex adopts the planar configuration, characterized by a Cu−Cu separation of 3.34–3.39 Å (structure 2 in Figure 5c). At the higher Si/Al ratios, about 50–60 % of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes adopts the bent configuration, with a Cu−Cu distance of 3.03-3.06 Å (structure 3 in Figure 5c). These results clearly show an effect of the Si/Al ratio of the Cu-CHA material on the structure of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex.

The bent and planar configurations are possibly important for the reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. Steady-state kinetics and operando XAS measurements⁷⁴ show that, at low Al content (Si/Al >12), the NH₃-SCR activity increases with the Al content, which is ascribed to easier paring of the (NH₃)₂Cu^I complexes. In catalysts with higher Al content, the reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex becomes more difficult, possibly due to the presence of more than one co-caged NH₄⁺. This would be an explanation for the well known inhibition of the NH₃-SCR rate by ammonia.^{75, 76} This behavior parallels the presence of the bent and planar configurations of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, and it could therefore be speculated that the more difficult reduction is related to the change in structure of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex from the bent to the planar configuration.

5 Reactivity of the [(NH₃)₄Cu₂O₂]²⁺-Peroxo Complex

The NH₃-SCR reaction cycle consists of alternating oxidation and reduction steps, with the oxidation of the linear (NH₃)₂Cu^I complex to form the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex as the oxidation step. The NH₃-SCR cycle is then closed by a reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex to the linear (NH₃)₂Cu^I complex via reactions with NO and NH₃. Consequently, the low-temperature NH₃-SCR activity of Cu-CHA catalysts depends on the reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex towards NH₃ and NO.

Experimental confirmation of the reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex in a mixture of NO and NH₃ to the linear (NH₃)₂Cu^I is provided by in-situ XAS. Exposure of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex to a 1 : 1 mixture of NO and NH₃ at 200 °C results in a complete conversion to (NH₃)₂Cu^I._.²⁴ In the WT analysis of the corresponding EXAFS spectra, the sub-lobe around k=7 Å^-1 is absent after reduction, indicating that the reduction is accompanied by the cleavage of the Cu−Cu core in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. The reduction is most efficient when NO and NH₃ are present together, though some reduction of Cu also occurs in the presence of NO or NH₃ alone, albeit to a lesser extent.^{21, 24} Interestingly, the sub-lobe at k=7 Å⁻¹ also disappears when the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is exposed to NO or NH₃ separately at 200 °C, which implies that in both cases the Cu−Cu core of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is cleaved, or at least, the bonding of the Cu-ions to the peroxo-group is modified to result in a less ordered Cu−Cu configuration.

The reaction of NH₃ with [(NH₃)₄Cu₂O₂]²⁺-peroxo complex in the absence of NO leads to a mixed-valence state, containing a Cu^I fraction of ca. 58 % total Cu, as estimated by XANES Linear Combination Fit analysis, similar to the partial reduction by NH₃ of Cu-nitrate.¹⁵ This indicates that reaction of NH₃ with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex does not result in the complete reduction of the Cu to the linear (NH₃)₂Cu^I complex, supporting the conclusion that NO is essential for the reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. According to EXAFS, the Cu species formed are most likely the linear (NH₃)₂Cu^I complex, and a [Cu^II(NH₃)₃(X)]⁺ species, similar to a [Cu^II(NH₃)₃(OH)]⁺ species.^{44, 57}

Upon reaction with NO, the XANES region of the spectra show predominantly a Cu^I state, characterized by the absence of a pre-edge peak at ∼ 8977 eV, stemming from the 1 s→3d transition in d⁹ Cu^II centres, and a prominent 1 s→4p peak at ∼8982 eV, indicative for the formation of the linear (NH₃)₂Cu^I complex.²⁴ This indicates an effective reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex and means that the reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex essentially takes place through a reaction with NO, and that at least a part of the NH₃ ligands are retained. A reduction of Cu by NO has been proposed earlier, under the formation of NO⁺ or NO₂⁷⁷ as intermediates. In more recently proposed reaction mechanisms, the reaction of NO leads to the formation of H₂NNO and HONO intermediates.^{9, 10, 76} We note, that the reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex corresponds to exposure of the (NH₃)₂Cu^I complex to O₂ and NO at 200 °C in consecutive steps; if the (NH₃)₂Cu^I complex is exposed to O₂ and NO simultaneously at 200 °C, a Cu-nitrate species is formed, which cannot be reduced by NO.¹⁵ This indicates a higher reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex for NO, as compared to Cu-nitrate, and therefore Cu-nitrate is less likely to be an intermediate in the low-tempeature NH₃-SCR reaction.

It has been attempted to exploit the reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with NO as a basis to quantify the amount of [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes in a Cu-CHA catalyst via titration methods. Such an approach would provide characterization method to monitor the ability of a Cu-CHA material to form the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. An easy way to do this would be to prepare the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex via oxidation of the reduced (NH₃)₂Cu^I complex with O₂,^{11, 27, 30} followed by measurement of the NO consumption under controlled conditions. However, the temperature of NO exposure in such measurements causes a change in the measured NO uptake per Cu^II ion from 0.77 at 150 °C to 1.04 at 250 °C.²⁷ This is a problematic result, since the amount of [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes does not depend on the temperature at which the titration is done, and therefore, the measured amount should not depend on the chosen temperature for NO exposure. This means that a measurement of the amount of [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes based on a measured NO consumption may be not conclusive.

Furthermore, the measured NO uptake of about one molecule NO per Cu^II[27] does not match the expected uptake that would be required for a complete reduction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex to 2 (NH₃)₂Cu^I, which would require two NO molecules per Cu^II, according to the stoichiometry of the NH₃-SCR reaction cycle. An explanation is that the reaction of NO with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex stops at some point, due to the lack of NH₃. However, the consumption of 1 NO molecule per Cu has also been measured in a complete reduction of the Cu, resulting in the formation of one N₂ molecule per reduced Cu ion.^{16, 19, 20, 32} Nasello et al. propose that this stoichiometry follows from a reaction cycle with the following reduction (RHC) and oxidation (OHC) half cycles:¹⁶

$$ (6)

(7)

We note that the sum of these cycles results in the overall stoichiometry of the NH₃-SCR reaction. The reactions explain the experimentally observed one NO molecule consumed and one N₂ molecule formed per reduced Cu ion. The 2 Z-Cu^IIOH take here the role of the reducible Cu, and should be kinetically equivalent to the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex in such measurements. However, these half cycles are inconsistent with the experimental observation of the formation of a [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. Furthermore, the oxidation step according to Eq. 3 requires some water, but it is known that the oxidation of Cu^I and NH₃-SCR also takes place in a dry atmosphere in the range 150–300 °C.^{10, 11, 15, 78} Therefore, it is not likely that H₂O acts as a reactant in the NH₃-SCR reaction. In fact, H₂O seems to inhibit the NH₃-SCR reaction in this temperature range.⁷⁸

To account for the formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, we can adjust Eqs. 3 and 4 as follows. First, we observe that the O₂ bridge in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex contains 2 O⁻ ions, given that the Cu ions are Cu^II ions, yielding the overall 2+ charge on the complex as required by charge neutrality for the zeolite system. This negative charge would then correspond to that of the OH⁻ ions in the two Z-Cu^IIOH groups in Eqs. 3 and 4. Therefore, Eqs. 3 and 4 could be reformulated as follows (omitting the NH₃ ligands for clarity):

$$ (8)

(9)

Eq. 5 then represents the oxidation of a pair of (NH₃)₂Cu^I to the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. No water is needed for the oxidation, in agreement with the observation that the oxidation and formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex also occurs in dry environments.²⁴ Adding Eqs. 8 and 9 results in the correct stoichiometry for NH₃-SCR (see Eq. 1). However, an oxygen atom is formed, which then needs to react somewhere in the catalytic cycle, without further consumption of NO, to keep the NO:Cu ratio to 1 : 1. Following this reasoning, some O₂ should be formed during the reduction, though there exists no experimental evidence for such a O₂ formation. The fundamental difficulty of the reaction stoichiometry as discussed above originates from the fact that the oxygen atoms react separately in the NH₃-SCR reaction, thus at some point leaving a single O atom. If that O atom reacts with the reactants NO, the NO:Cu ratio becomes 2 : 1, otherwise we obtain a NO:Cu ratio of 1 : 1.

The complications and the lack of understanding of the precise reactions of NO with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex mean that, at present, NO titrations are not yet reliable enough to measure the amount of [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes. An alternative way to measure the amount of reducible Cu^II is by CO oxidation at 200 °C.^{26, 79-81} This circumvents the complications of possible side reactions with NH₃ present in the catalysts. The amount of formed CO₂ measures the amount of reducible Cu^II species. Since the reaction requires multinuclear Cu-complexes, this method provides a reliable measurement of the amount of [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes in Cu-CHA catalysts.

6 First-Principles Reaction Kinetics for NH₃-SCR

The knowledge about the formation and reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex forms the basis for NH₃-SCR reaction cycles guided by first-principles (DFT) calculations, using the mean-field approach.^{9, 10} According to the current understanding, the only reaction step capable of oxidation of the (NH₃)₂Cu^I complexes at around 200 °C is the oxidation by O₂, resulting in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. Therefore, the NH₃-SCR activity is determined by the reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with NO and NH₃. By taking into account the entropy effects, the free energy profiles of the reaction cycle and activation barriers (ΔG^≠) are calculated, which correspond to the rate constants for the individual reaction steps, according to the transition state theory.^{82, 83} The advantage of such models is that the level of detail allows for relating observed kinetic effects to precise structures and reaction steps in the reaction cycle. The challenge, however, is to validate such detailed models with experimental observations, and even if the model agrees with experimental kinetics, other pathways may still be possible.

Figure 6 shows two possible reaction cycles, based on DFT calculations, for the entire NH₃-SCR reaction.¹⁰ The cycles are based on a reaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with NO. In cycle I, the reaction of the NO molecule takes place on one of the Cu ions in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex; in cycle II, the NO reacts with the bridging peroxo-group. This results in different reaction intermediates, even though the overall reaction paths are similar. The reaction with NO leads to HONO and H₂NNO intermediates, that are released from the Cu complexes to react to the products N₂ and H₂O on the acidic Brønsted sites in the zeolite.⁹ This model implies, that the overall NH₃-SCR reaction requires both a Cu ion for the initial reaction steps, and a Brønsted acid site to form the N₂ and H₂O reaction by decomposition of the HONO and H₂NNO intermediates. In both cycles, the reaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with NH₃ results in an extra NH₃ ligand, rendering the complex inactive for NH₃-SCR. The latter reaction accounts for the known inhibition of the NH₃-SCR reaction by NH₃^{75, 76} in this model.

Figure 6

Proposed reaction cycles for the NH₃-SCR reaction over Cu-CHA catalysts, based on the reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. Cycle I: reaction of NO on a Cu-center. Cycle II: reaction of NO on the oxygen bridge in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. Reproduced from Ref. [10]. Copyright © 2021 Y. Feng et al. Published by American Chemical Society. This publication is licensed under CC-BY 4.0.

A further analysis of the kinetics of the reaction cycles in Figure 6 indicates that cycle I is about 10 times faster than cycle II below 250 °C,¹⁰ which suggests that cycle I gives the better description of the NH₃-SCR reaction at low temperatures. A comparison of the calculated reaction orders for NO, NH₃, and O₂, and the calculated activation energies based on these reaction cycles is given in Table 1.¹⁰ The calculated activation energies of 0.75 eV (72 kJ/mol) for cycle I, with the inclusion of the NH₃ inhibition step (step 15), agrees well with the experimentally determined values.^{15-17, 28, 75, 77} Likewise, the corresponding reaction orders agree well with the experimental values as well, though the experimentally determined reaction order for NO of about 0.7 is slightly lower than the calculated value of 0.95..^{10, 75} It is noted that omission of the NH₃-inhibition step in cycle I results in an activation energy of 0.31 eV (30 kJ/mol), indicating that the experimentally observed activation energy is largely determined by the NH₃-inhibition step, even though that reaction step is a side reaction to the main NH₃-SCR reaction cycle. The calculated activation energies based on cycle II clearly deviate from the experimental results, which is consistent with the conclusion that this reaction cycle does not significantly contribute to the overall reaction rate below 250 °C.¹⁰

Table 1. Experimental and Calculated Apparent Activation Energy and Reaction Orders for NH₃-SCR over Cu-CHA (1.6 wt% Cu, Si/Al=6.7). From Ref. [10].

	exp.	cycle I	cycle I w/o NH3 inhibition (step 15)	cycle II	cycle II w/o NH3 inhibition (step 26)
apparent Eact	0.62/0.89	0.75	0.31	1.06	0.05
order NH3	−0.23	−0.22	0.05	−0.76	0.00
order NO	0.73	0.95	0.92	1.00	0.99
order O2	0.25	0.23	0.34	0.15	0.58

We note that the reaction cycles I and II do not include a dissociation of the Cu-pair upon interaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with NH₃, as has been determined experimentally²⁴ (see Sect. 5), but merely state that the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex becomes inactive for NH₃-SCR upon reaction with NH₃. A dissociation of the Cu-pair would also result in an inactive intermediate, and therefore, the two descriptions will result in similar kinetic behavior, as discussed in a simplified version of the kinetic model.⁸⁴

To determine the critical reaction steps for the NH₃-SCR rate, a degree of rate control analysis has been applied to reaction cycle I, for the temperature range 100–250 °C; the results are shown in Figure 7. According to this analysis, the reaction of NO with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex (step 2) is rate determining below 150 °C, as the availability of [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is determined by the NH₃-inhibition equilibrium (step 15). At higher temperatures, a reaction step of NO with a [(NH₃)₄Cu₂OH₂]²⁺-complex (step 8) and the first reaction step with NH₃ (step 3) start to contribute to the overall reaction rate; at 250 °C, step 2, 3, and 8 contribute in about equal amounts to the overall rate. As steps 3 and 8 are considered to be a part of the reduction half cycle, this result corroborates the observation that the overall reaction rate for NH₃-SCR in the temperature range 200–250 °C is determined by both the oxidation and reduction steps, without a clear single rate determining step.^{16, 17, 20, 32} In fact, step 8 corresponds to the reduction of 2 Z-CuOH groups, as mentioned in Eq. 6.¹⁶ The corresponding fractions of Cu^I and Cu^II derived from this model agree well with experimental observation.^{10, 17} The detailed knowledge of the influence of the individual reaction steps also allows for construction of simpler kinetic models, that preserve the relation to the underlying chemistry and result in a good approximation of the reaction rates.^{10, 84}

Figure 7

Degree of rate control analysis for cycle I. Reproduced from Ref. [10]. Copyright © 2021 Y. Feng et al. Published by American Chemical Society. This publication is licensed under CC-BY 4.0.

Other interesting aspects of the reaction cycles in Figure 6 are the different roles of NH₃, and the formation of N₂O. Four different roles for NH₃, namely (1) ligand NH₃, bound to the Cu ions, (2) inhibiting NH₃, adsorbing on the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex (3) reactant NH₃, and (4) NH₃ as NH₄⁺ on the Brønsted acid sites are outlined.¹⁰ In order to construct precise kinetic models for NH₃-SCR, these roles must be accounted for, implying that models based on a single total NH₃-coverage or NH₃-content must be expected to have limited validity. In the models, the reactant NH₃ is the NH₃ in the gas phase, entered via the partial pressure.

Cycle I includes a reaction path for the formation of N₂O. The formation of N₂O has been associated to a decomposition of ammonium nitrate in the pores of the CHA zeolite.^{6, 85} However, the decomposition of ammonium nitrate has an activation barrier of about 1.5-1.8 eV,^86-88 and more favorable energy landscapes are found for N₂O formation as a side reaction.⁸⁶ Cycle I proposes that the formation of N₂O occurs in parallel to the first H₂NNO intermediate in the NH₃-SCR cycle (steps 5 and 16 in cycle I), without the formation of an ammonium nitrate species. The energy barrier for this step is 0.40 eV,¹⁰ significantly lower than the barrier for the decomposition of ammonium nitrate. In contrast to the much faster NH₃-SCR reaction, the N₂O formation does not require a Brønsted acid site to proceed.¹⁰ This means that, according to this model, the selectivity to N₂O is determined by the availability of Brønsted acid sites, and catalysts with a high Cu loading, not leaving sufficient Brønsted acid sites for the decomposition of the H₂NNO and HONO intermediates in the NH₃-SCR cycle, are therefore expected to show a higher selectivity for N₂O. Such a trend has been observed experimentally,⁸⁹ though a quantitative agreement still needs to be verified. Furthermore, because N₂O formation does not occur in reaction cycle II, this is a further indication that the NH₃-SCR reaction follows cycle I in the temperature range 150–300 °C.¹⁰

It should be noted that reaction cycle I assumes the presence of two Cu^I in a CHA-cage. However, the presence of Cu-pairs could limit the reaction, in particular at low Cu content.^{16, 30, 32} Assuming the presence of a pair overestimates the rate for the NH₃-SCR reaction. This seems to be corroborated by the fact that the calculated turnover frequency for the NH₃-SCR reaction of about 0.1 s⁻¹, based on cycle I¹⁰ is clearly higher than the TOF for the reduction of the Cu^II – this presumably is the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex – reported by Daya et al.²⁷ Furthermore, it is assumed that a Brønsted acid site for the decomposition of the H₂NNO and HONO intermediates is readily available. Particularly at higher Cu loadings, this is not necessarily true. According to cycle I, a lack of Brønsted sites would result in a lower NH₃-SCR rate. In fact, this might be the reason for the lower experimental value for the reaction order of NO, as compared to the calculated value based on cycle I. Finally, the NO consumption in cycle I (and cycle II) clearly is 2 per Cu ion, which does not match the value observed in the measurement of the reduction half cycle as explained in Sect. 5. However, because the NH₃-SCR reaction rate is largely determined by the oxidation half cycle,¹⁰ in particular below 200 °C, a change in the stoichiometry of the reduction does not necessarily result in a change of the rate of the entire NH₃-SCR cycle.

7 Deactivation of Cu-CHA Catalysts by SO₂

A drawback of Cu-CHA catalysts for NH₃-SCR is their sensitivity towards SO₂. Prolonged exposure to SO₂ can cause severe deactivation in the low-temperature range (below 350 °C). In practice, this means that application of Cu-CHA zeolites in exhaust systems requires the use of ultra-low sulfur diesel (ULSD), which contain less than 10 (Europe) or 15 (US) ppm sulfur on weight basis, to minimize the partial pressure of SO₂ in the exhaust gas. Generally, a Cu-CHA catalyst loses most of the activity below 350 °C upon exposure to SO₂, which to a large extent can be recovered by heating to 500–550 °C (see Figure 8, left panel). However, the deactivation occurs rapidly upon SO₂ exposure, but stabilizes after the loss of about 90 % of the original activity (see Figure 8, right panel); the catalysts never loose their low-temperature activity completely. Regeneration by heating then reinstates most of the lost activity, but, typically 10–30 % of the original activity cannot be recovered this way.^{90, 91}

Figure 8

Left panel: Measured NOx conversions over Cu-CHA catalyst (2.8 wt% Cu, Si/Al=14.7) in the fresh state, after exposure to SO₂ (100 ppm SO₂, 10 % O₂, bal. N₂ at 550 °C), and after regeneration (heating in 500 ppm NO/533 ppm NH₃/5 % H₂O/10 % O₂/bal. N₂ at 550 °C for 4.5 h). From Ref. [90]. Right panel: measured S/Cu ratio and deactivation with time of exposure to 50 ppm SO₂ at 200 °C. From Ref. [91].

The impact of SO₂ on the NH₃-SCR activity implies that SO₂ interacts with the Cu-CHA catalyst under NH₃-SCR conditions. Figure 9 shows how SO₂ affects the in-situ Cu−K edge XANES spectra of different framework bound and NH₃-solvated Cu^I and Cu^II -species in a Cu-CHA catalyst with 3.2 wt% Cu and Si/Al ratio of 6.7.¹¹ While the spectra for Cu^I and framework bound Cu^II only change slightly upon SO₂ exposure at 200 °C, the spectra for the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex and Cu^II in the presence of NH₃ show significant changes, indicating that SO₂ reacts with Cu^II in the presence of NH₃. In fact, the XANES spectra for the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex indicate a reaction with SO₂, in which the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex breaks up and the Cu is partially reduced to the linear (NH₃)₂Cu^I complex. This formation of (NH₃)₂Cu^I is further corroborated by the lower 1^st shell coordination around R=1.5 Å, as such a reaction implies a reduction in coordination from 4 to 2. The total uptake of SO₂, as determined by X-ray Adsorbate Quantification (XAQ)⁹² and temperature programmed desorption of SO₂ (SO₂-TPD), is also clearly enhanced for Cu^II species in the presence of NH₃ (Figure 9 right panel). These results indicate that the interaction of SO₂ with the Cu in the Cu-CHA catalysts preferably proceeds via a reaction with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. The observation that poisoning by SO₂ is enhanced in the presence of an NH₃-SCR feed gas (NH₃/NO/O₂/H₂O), as compared to poisoning by SO₂ in the presence of O₂ and H₂O only,⁹³ provides further indirect evidence for the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex as a critical component for deactivation by SO₂.

Figure 9

Left panel: In-situ Cu−K edge XANES and corresponding FT EXAFS upon exposure of Cu-CHA (3.2 wt% Cu, Si/Al=6.7) to 400 ppm SO₂ for framework-bound Cu^I_, (NH₃)₂Cu^I, (NH₃)₂Cu^I heated to 500 °C, framework-bound Cu^II, the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, and framework-bound Cu^II exposed to NH₃. Right panel: Measured sulfur uptake by X-ray Adsorbate Quantification and SO₂-TPD from the same Cu-species. From Ref. [11].

The fact that SO₂ reacts with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex leads to the hypothesis that deactivation by SO₂ is caused by the formation Cu−S species that are not active for NH₃-SCR. An analysis of the Cu−K edge XANES data indicate the presence of such a sulfated Cu-compound (Figure 10, left panel).¹² The shape of the S K-edge XANES and S Kα XES spectra after interaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with SO₂ is typical for S^VI in tetrahedral coordination, confirming the previous ex-situ results (Figure 10, right panels).^{94, 95} This means that the Cu−S compound formed in the reaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with SO₂ is a sulfate-like compound. The formation of a sulfate-like compound is corroborated by the Cu−K-edge XANES and corresponding EXAFS.^{11, 12} The Cu K-edge XANES features of the Cu−S compound are similar to those of Cu²⁺ in a quasi-square-planar geometry and a coordination to N or O atoms.

Figure 10

Left panel: In-situ XANES at the S−K edge upon exposure of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex to 100 ppm SO₂ at 200 °C. Right panel: Comparison of the S−K edge in XANES and XES S Kα emission lines of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex exposed to SO₂ with reference compounds. Reproduced from Ref. [12] with permission from the Royal Society of Chemistry.

A fit of the corresponding EXAFS points to a structure with the two O and two NH₃ ligands situated at slightly different distances (1.94 and 2.09 Å), together with a contribution of S atom, most probably bound to Cu through oxygen ligands, at 2.58 Å, in line with the formation of Cu-sulfate-like species. From the observed formation of Cu^I and Cu^II compounds in the reaction with SO₂, a two-step reaction was proposed. In the first step a framework-bound Cu^I and a sulfur based intermediate, presumably SO₃, are formed, which is then followed by a reaction of that sulfur based intermediate with a second [(NH₃)₄Cu₂O₂]²⁺-peroxo complex to form a Cu-sulfate species, and the (NH₃)₂Cu^I complex.¹² Overall, these experimental XAS and XES data provide evidence that [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is the most reactive intermediate in the NH₃-SCR cycle towards SO₂. This reaction then leads to a direct interruption of the NH₃-SCR cycle, and the accumulation of sulfur in the catalyst, resulting in the deactivation.

Another approach to understand the interaction of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex with SO₂ is to construct a reaction model using density functional theory (DFT). Using this approach, it has been proposed,¹³ that SO₂ attaches to the oxygen-bridge in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, in agreement with the spectroscopic data,^{11, 12} however, arriving at a different mechanism for the deactivation of the Cu-CHA. In the scheme in Ref [13], the Cu^I species does not form immediately in a reaction with SO₂, but would still require a further reaction step with NO, in contrast to the observation in XAS, that Cu^I species appear in the absence of NO.¹² Eventually ammonium sulfate is formed, via a sulfuric acid intermediate, and Cu-sulfate like species. The presence of ammonium sulfate then influences the ability of the (NH₃)₂Cu^I complexes to form the Cu pairs and formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complexes. In fact, it is shown that the presence of a single ammonium-sulfate unit in a zeolite cage actually stabilizes the formation of Cu-pairs, and that the destabilization of Cu-pairs first occurs when two ammonium sulfate units are present in the cage.¹³ This scenario would imply that Cu-CHA catalysts can tolerate SO₂ below a certain critical uptake of sulfur. This, however, contradicts the observation that the deactivation is fast in the initial phases of SO₂ exposure, and that the observed activity is decreased by 90 % at S/Cu ratios well below 1 (see Figure 8, right panel).⁹¹

Because the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is solvated by NH₃-ligands, it cannot directly be related to a specific site in the zeolite, such as the Z₂-Cu^II site in the 6mr-rings, or a Z-Cu^IIOH site, which are the most common locations for a Cu^II ion in the zeolite. Consequently, the effect of SO₂ on the low-temperature NH₃-SCR activity is not directly determined by the location of the Cu ions in the zeolite. Moreover, the Z₂-Cu^II and Z-Cu^IIOH sites represent framework-bound Cu^II species, which are not very reactive towards SO₂ (see Figure 9).^{11, 12} This suggests that, in principle, the deactivation by SO₂ is not very dependent on the framework locations of the Cu ions in the zeolite. Nevertheless, a number of studies rationalize experimental observations in terms of the reactivity of Z₂-Cu^II and Z-Cu^IIOH sites towards SO₂.^96-101 Experimental evidence for a different interaction of SO₂ with these Cu species is provided by FTIR of antisymmetric [TO₄] vibrations of the zeolite framework in the 900–1000 cm⁻¹ range. The strong electrostatic interaction of the framework-bound Z₂-Cu^II and Z-Cu^IIOH perturbs the antisymmetric [TO₄] vibration of the zeolite framework. When Z₂-Cu^II and Z-Cu^IIOH become solvated by NH₃, the perturbation is weakened, resulting in negative bands at 900 and 950 cm⁻¹ in the FTIR spectrum, corresponding to the solvation of the Z₂-Cu^II and Z-Cu^IIOH, respectively.^{96, 97, 102, 103} Figure 11 shows the effect of SO₂ on the solvation process for the Z₂-Cu^II and Z-Cu^IIOH species, by monitoring the FTIR bands at 900 and 950 cm⁻¹ during exposure to NH₃ in the absence and presence of SO₂. When SO₂ is present during the NH₃ exposure, the measured intensity ratio I₉₅₀/I₉₀₀ is lower than that measured in the absence of SO₂ (left and middle panels in Figure 11). This indicates that the effect of SO₂ is different for the Z-Cu^IIOH and Z₂-Cu^II sites.

Figure 11

FTIR of the TO₄-framework vibrations in Cu-CHA (Si/Al ratio 30, 1.2 wt% Cu) upon exposure to NH₃ + O₂ at 200 °C (left panel) or NH₃+SO₂+O₂ at 200 °C (middle panel), and SO₂-TPD from Cu-CHA with Si/Al ratio 6 and 30 (right panel). Reprinted with permission from Ref. [97] Copyright © 2018, American Chemical Society.

A different interaction for Z₂-Cu^II and Z-Cu^IIOH is further supported by temperature-programmed desorption of SO₂ (SO₂-TPD). In a Cu-CHA catalyst with Si/Al=30, a larger fraction of the SO₂ desorbs at around 600 °C, as compared to a Cu-CHA catalyst with Si/Al=6, as shown in the right panel in Figure 11. Since a catalyst with a high Si/Al ratio is expected to contain more Z-Cu^IIOH,¹⁷ this indicates a stronger interaction of SO₂ with a catalyst that contains more Z-Cu^IIOH. The total amount of SO₂ observed in these TPD measurements are 563 μmol/g for the 3.58 wt% Cu-CHA (Si/Al=6) catalyst, and 204 μmol/g for the 1.20 wt% Cu-CHA (Si/Al=30) catalyst;⁹⁷ these values correspond to S/Cu ratios of 1.01 and 1.08, respectively. This suggests that the total uptake of SO₂ is limited and determined by the amount of Cu in the zeolite, in full agreement with the formation of a Cu-sulfate-like compound, as determined from XAS measurements.¹² However, the two distinct features in SO₂-TPD also indicate that different Cu-sulfate-like species are formed, dependent on the Cu content and Si/Al ratio of the zeolite. The formation of excessive amounts of ammonium sulfate can then be excluded, as this would lead to significantly higher amounts of SO₂ in the TPD measurement.

To reconcile the observed differences in the interaction of SO₂ with Z-Cu^IIOH and Z₂-Cu^II with the observation that the interaction of SO₂ takes place via an interaction with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, the formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex from Z-Cu^IIOH and Z₂-Cu^II must be considered. In principle, the formation of [(NH₃)₄Cu₂O₂]²⁺-peroxo complex depends on the mobility and pair formation of the (NH₃)₂Cu^I complexes. Experiments indicate, that both Z₂-Cu^II and Z-Cu^IIOH sites are reduced and mobilized as (NH₃)₂Cu^I complexes upon reduction in a NO/NH₃ mixture at 200 °C.^{56, 104} The question remains whether and how the mobility and reactivity of (NH₃)₂Cu^I complexes are affected by the Al content, and the corresponding negative charge density, in the zeolite. Furthermore, Z-Cu^IIOH sites can be mobilized as [(NH₃)₃Cu^II(OH)]⁺ in the presence of NH₃, which can diffuse between the zeolite cages in the CHA structure. In contrast, inter-cage diffusion is not possible for [(NH₃)₄Cu^II]²⁺ formed from Z₂-Cu^II, due to the electrostatic tethering of the divalent complex.^{26, 80} This different behavior of Z-Cu^IIOH and Z₂-Cu^II provides a way how the corresponding reduction and pair formation of (NH₃)₂Cu^I complexes can be different. The Cu^I-pairs formed from the more mobile Z-Cu^IIOH may be located further away from their Al anchor points in the zeolite structure.

In conclusion, it is certain that the interaction of SO₂ with Cu-CHA catalyst for NH₃-SCR proceeds via the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. That reaction leads to a partial reduction of the Cu to Cu^I, and the oxidation of SO₂ to a Cu-sulfate-like species, while the formation of ammonium sulfate seems less likely. The S atom in the Cu-sulfate-like species is in an S^VI oxidation state, which is bound to the Cu-ions via an oxygen atom, as expected for a Cu-sulfate-like species. The precise identification of all compounds involved in the uptake of sulfur, and the mechanism for the observed deactivation, however, have not been solved yet. Because a small amount of SO₂, corresponding to S/Cu ratios of 0.4-1 causes a loss of 80–90 % of the low-temperature activity for NH₃-SCR, the most probable mechanism for the deactivation by SO₂ is a disturbance of the NH₃-SCR reaction cycle via the interaction of SO₂ with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex.

8 Final Remarks

The basis of the current understanding of the Cu-CHA catalysts for NH₃-SCR in the temperature range 150–300 °C is that the first step is the adsorption of molecular oxygen on a pair of mobile (NH₃)₂Cu^I complexes.^{18, 30, 34} The key role of this step provides insight in how catalyst properties, such as the Cu content, the Al content and Al distribution can influence the reaction. The chemical properties [(NH₃)₄Cu₂O₂]²⁺-peroxo complex that is formed determine to a high degree the catalytic behavior of Cu-CHA catalysts in NH₃-SCR. The NH₃-SCR activity is governed by the reactivity of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex towards NO (see Sect. 5),²⁴ and the application of this idea has led to a successful kinetic model of the NH₃-SCR reaction based on the formulation of a reaction pathway at the molecular level (see Sect. 6).^{9, 10} The negative reaction order in NH₃ for NH₃-SCR (NH₃-posioning) results from a reaction of NH₃ with [(NH₃)₄Cu₂O₂]²⁺-peroxo complex¹⁰ (see Sect. 6), which leads to a dissociation of the Cu-pair in the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex (see Sect. 5).²⁴ Finally, the deactivation by SO₂ is a consequence of a reaction of SO₂ with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, under the formation of Cu-sulfate-like compounds (see Sect. 7).^{11, 12}

There are also some remaining questions regarding impact of the chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex on the performance of Cu-CHA catalysts for NH₃-SCR. First, the ‘bent’ and ‘planar’ configurations of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex seems to result in a different rate of the reduction half cycle (see Sect. 4). It remains unclear if this is really the case, how it works, and how the configuration of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex is affected by the structure of the zeolite, Si/Al ratio and Cu content. Second, the reaction cycles based on the formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex imply a 2 : 1 NO:Cu ratio in the reduction half cycle of the NH₃-SCR reaction,^{9, 10} whereas separate measurements of the oxidation and reduction half-cycles indicate a 1 : 1 NO:Cu ratio.^{16, 19, 20, 32} This could mean that the reaction cycles proposed so far are not entirely correct, even though they result in correct reaction orders, and activation energies. Third, the deactivation by SO₂ results from a reaction of SO₂ with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, under the formation of a Cu-sulfate-like species, the sulfur atom in the S^VI oxidation state. However, this resulting Cu-sulfate-like species has not been identified exactly yet.¹² Furthermore, a DFT based model for deactivation by SO₂ suggests that the deactivation is due to accumulation of some ammonium sulfate in the zeolite cages,¹³ which is not consistent with the observed deactivation behavior.⁹¹

Other aspects of the chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex that have not yet been considered are the thermal stability of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex, and the possible impact of water on the reactions with the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. With respect to the thermal stability, we can expect that the NH₃ ligands are lost at higher temperatures, similar to the decomposition of the linear (NH₃)₂Cu^I complex, which occurs around 250–300 °C.^{57, 105} The well-known decrease in NOx conversion around 300 °C (“seagull” effect) is often related to the decomposition of the (NH₃)₂Cu^I complexes, resulting in the loss of their mobility,^{18, 58} such that the formation of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex becomes more difficult. However, it can be expected that the thermal stability of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex plays a similar role. In addition, the reactivity of the oxygen bridge may change at higher temperatures, and this may affect the rate of the NH₃-SCR reaction via a change of the reduction half cycle.

The impact of water on the chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex has so far not been considered. It is known that the presence of water affects the kinetics of the NH₃-SCR reaction. For example, the activation energy of the reduction half cycle is 20 kJ/mol lower in the presence of water,¹⁰⁶ which may due to a change in the chemistry of the [(NH₃)₄Cu₂O₂]²⁺-peroxo complex. Moreover, below 250 °C, Cu^II becomes solvated by water molecules,^{107, 108} introducing another type of mobility in the system. In view of the recent developments for ammonia-powered and hydrogen-powered internal combustion engines (HICE), which are possible technologies to reduce greenhouse gas emissions in the transportation sector, the impact of water on the performance of NH₃-SCR catalysts will become an important issue. In a hydrogen-powered combustion engine, the water content in the exhaust will typically be well above 10 %, dependent on the air/fuel ratio, and becomes 34 % at the stoichiometric air/fuel ratio.^{109, 110} Hydrogen-powered engines produce on the order of 500–5000 ppm NOx, primarily due to the higher temperature in the combustion, as compared to conventional combustion engines.^{110, 111} NOx emissions from hydrogen-powered combustion engines can be effectively reduced by conventional Cu-CHA catalysts.¹¹⁰ Understanding the impact of a high concentration of water will therefore be important for a successful implementation of Cu-CHA catalysts for NOx abatement in combustion engines based on non-fossil fuels.

Acknowledgments

Support from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement  No. 955839 (CHASS) is gratefully acknowledged. HG acknowledges financial support from the Swedish Energy Agency (47110-1 and 52689-1). GB and EB acknowledge support from the Project CH4.0 under the MUR program “Dipartimenti di Eccellenza 2023–2027” (CUP: D13C22003520001).

Conflict of Interests

The authors declare no conflict of interest.

Biographical Information

Ton V.W. Janssens is Senior Scientist at Umicore Denmark ApS in the Automotive Catalysis department, working on catalysts for the selective catalytic reduction of NOx diesel exhaust systems. Ton obtained his PhD in 1993 at the Eindhoven University of Technology, the Netherlands, in the field of surface science. He then moved to the University of California, Riverside, and the Fritz-Haber Institute, Berlin, as a postdoc, after which he joined Topsoe (Denmark) in 1998. At Topsoe he initially worked on catalysts for syngas conversion, and zeolite catalysis. He moved to the Automotive Catalysis department in 2012, and joined Umicore in 2017.

Biographical Information

Elisa Borfecchia obtained her PhD degree in 2013 at the University of Turin. During her PhD and afterward, her research centred on characterizing functional materials by X-ray spectroscopy. From 2016, she worked at Haldor Topsøe A/S and she was subsequently awarded a Marie Curie Fellowship to join the Catalysis group at Oslo University. From 2019, she has been assistant professor in Physical Chemistry at the University of Turin, where she is currently Associate Professor. She authored >90 peer-reviewed papers and performed >60 experiments at synchrotrons, aimed at disclosing local structure and reactivity of metal centres in heterogeneous catalysts.

Biographical Information

Kirill A. Lomachenko is responsible for the chemistry research programme at BM23 and ID24 XAS beamlines of the European Synchrotron Radiation Facility (France). He completed his PhD under the joint supervision of Profs. Carlo Lamberti (University of Turin, Italy) and Alexander Soldatov (Southern Federal University, Russia) in the intersection of physics, chemistry, and materials science. His main field of research is the application of XAS and XES spectroscopies and related characterization techniques to establish the structure-property relationships of novel materials for catalysis, energy storage, and gas sorption.

Biographical Information

Henrik Grönbeck received his PhD in Physics from Göteborg University in 1996 with a thesis on electronic and structural properties of metal clusters. He joined Chalmers in 2001 after working at Ericsson Microwave Systems and a postdoc position at IBM Research Zurich. Grönbeck was promoted to professor in 2012 and is presently the head of the Division of Chemical Physics. His group is working within computational catalysis and surface science, and one main research direction is to develop kinetic models based on density functional theory calculations.

Biographical Information

Gloria Berlier received her PhD in Chemical Science from the University of Turin in 2001 with a thesis on the structure and reactivity of iron clusters in zeolites. She worked as post-doc at the DFRL of the Royal Institution of GB in London and at the Department of Chemistry of University of Turin. Since 2021 she is Full Professor in Physical Chemistry. She works at the characterization of heterogeneous catalysts, focusing on the structure and reactivity of surface sites. Her experimental approach involves in situ and operando optical spectroscopies, electronic microscopy, gas-volumetric and structural techniques.