## Introduction

Increasingly stringent mobile source emission regulations have been pursued around the world to tackle environmental pollution. Nitrogen oxides (NOx) are inevitable gaseous pollutants emitted from internal combustion engines. Selective catalytic reduction of NOx with NH3 (NH3-SCR) is the most widely adopted technology for the removal of NOx from diesel engines1,2. The successful commercialization of Cu-SSZ-13 as an NH3-SCR catalyst is a significant achievement for diesel engine exhaust post treatment3. In the past decade, numerous studies have endeavored to uncover the standard SCR (SSCR) reaction mechanism4,5,6,7, hydrothermal deactivation mechanism8,9,10,11, and SO2 poisoning deactivation mechanism12,13,14, and to develop economic and sustainable synthesis methods for Cu-SSZ-1315,16,17,18, bringing about continuous optimization of Cu-SSZ-13 for commercial SCR catalysts.

In actual application, a diesel oxidation catalyst (DOC) is utilized to oxidize carbon monoxide (CO) and hydrocarbons (HCs), accompanied by partial oxidation of NO to NO2. The formed NO2 can participate in the NH3-SCR process through the so-called “fast SCR” reaction (FSCR, reaction 1, consisting of reactions 2 and 3). It is generally believed that the deNOx efficiency of the FSCR reaction should be higher than that of SSCR (reaction 4) due to bypassing the oxidation of NO, which is usually the rate-limiting step in the SSCR reaction on V-based and Fe-zeolite catalysts19,20.

$$NO+N{O}_{2}+2N{H}_{3}\to 2{N}_{2}+3{H}_{2}O$$
(1)
$$2N{O}_{2}+2N{H}_{3}\to N{H}_{4}N{O}_{3}+{N}_{2}+{H}_{2}O$$
(2)
$$NO+N{H}_{4}N{O}_{3}\to {N}_{2}+N{O}_{2}+2{H}_{2}O$$
(3)
$$4NO+4N{H}_{3}+{O}_{2}\to 4{N}_{2}+6{H}_{2}O$$
(4)

However, there have been few studies reporting that NO2 measurably promotes the NH3-SCR efficiency over Cu-SSZ-13 catalytic systems. On the contrary, inhibition of NO conversion by NO2 was found over Al-rich Cu-SSZ-13 catalysts due to NH4NO3 formation, which is the so-called “abnormal fast NH3-SCR reaction”21. In our recent study, we found that the inhibiting effect of NO2 was closely related to Brønsted acid sites (BASs) and can be alleviated by hydrothermal aging due to the decrease in the number of BASs in Cu-SSZ-1322. Therefore, we speculated that NO2 reduction probably occurs at BASs. Also, we previously observed the reaction between NO and NH4NO3 occurring at BASs over the H-SSZ-13 catalyst23. Furthermore, Kubota et al. found that NO reacts with NH4NO3 more rapidly than NH4NO3 decomposition over H-AFX and H-CHA zeolites24,25. However, the situation in Cu-containing zeolites is more complicated. McEwen et al. found that four-fold-coordinated Cu(II) species dominate the Cu-SSZ-13 catalyst under FSCR conditions, which differs from the composition under SSCR conditions, where Cu(I) and Cu(II) species both exist26. Paolucci et al. investigated the oxidation process of Cu(I)(NH3)2 species by O2 and NO2. It was found that oxidation by NO2 occurred at isolated Cu sites, rather than at the Cu dimer sites required for O2 activation5. More recently, Liu et al. investigated the FSCR mechanism over the Cu-OH site on Cu-CHA zeolite and showed the important role of BASs in the FSCR reaction27. Therefore, it can be concluded that the FSCR reaction pathway over Cu-SSZ-13 is unique and different from other catalytic systems where NO2 accelerates SCR rates. The active sites as well as redox pathways may change over Cu-SSZ-13 in the presence of NO2. Compared to the relatively few studies on the FSCR reaction mechanism, researchers have conducted numerous experimental and theoretical studies to explore the SSCR reaction mechanism in the past decade. Thus, the SSCR mechanism has been relatively clear, in which dynamic binuclear Cu+ species are the primary active sites4,5,28. However, the influence of NO2 on the active Cu sites and the mechanism of the NO2-involved SCR reaction are barely discussed, and are worth exploring since NO and NO2 always coexist in actual applications.

In this study, the SCR reaction over the Cu-SSZ-13 catalyst in the presence of both NO and NO2 was studied by kinetic measurements. In situ X-ray absorption fine structure (XAFS) measurements were applied to reveal the state of copper species under SSCR (with only NO as NOx), FSCR (equal mixture of NO and NO2 as NOx) and NO2-SCR (only NO2 as NOx) reaction conditions. Density functional theory (DFT) calculations were conducted to identify the NO2-involved SCR reaction pathways. These results provide new insights into the role of NO2 in the NH3-SCR reaction and shed light on the actual application of Cu-SSZ-13 catalysts in the presence of both NO and NO2.

## Results and discussion

### Wavelet transform analysis of in situ EXAFS measurements

Further, we conducted in situ XAFS measurements on Cu-SSZ-13 samples to uncover the valence state and coordination of copper species under different conditions. Wavelet transform (WT) analysis of extended X-ray absorption fine structure (EXAFS) spectra is a powerful technique to resolve overlapping contributions from different neighbor atoms at close distances around the absorber. As shown in Fig. 3a, the pretreated sample shows a distinct first shell peak at (4.5 Å-1, 1.3 Å), which is associated with contributions from framework oxygen atoms. This result suggested that the copper species mainly exist as fw-Cu2+ species, which have high coordination numbers28. For the second shell sphere (R(Å) > 2 Å), two lobes, at (3.5 Å−1,2.8 Å) and (6.5 Å−1 3.3 Å), are well-resolved due to the different backscattering properties of various atoms, which strongly depend on the atomic number. The first lobe is assigned to the second-shell oxygen atom due to the low k value of oxygen atoms. The latter one is attributed to the signals from the Si or Al atoms of the framework. Although some studies attributed the latter lobe to the Cu-Cu contributions in oxygen-bridged Cu dimers38,39, we scarcely observed CuOx species in X-ray absorption near edge structure (XANES) and EXAFS profiles (Supplementary Fig. 12) and did not carry out the procedure of introducing O2 to NH3-treated Cu-SSZ-13 to form oxygen-bridged Cu dimers with four NH3 ligands. Therefore, we deduced that the lobe at 6.5 Å−1 is primarily derived from the framework Si or Al atoms in the second shell in this work. In fact, the copper species in Cu-SSZ-13 are initially in the solvated state as [Cu(H2O)n]2+ under ambient conditions, which weakens the interaction between copper species and the zeolite framework28,40. High-temperature treatment in O2/N2 removes the coordinated water molecules and oxidizes copper species to Cu2+. As a result, the copper species are in a high valence state and strongly interact with the zeolite framework through electrostatic forces.

After NO adsorption, Cu2+ ions are partially reduced, resulting in a slight decrease in the coordination numbers (CNs) of the first shell, denoted by the decrease and weakening of the colored area (Fig. 3b). The lobes resulting from the contributions of the second shell stretched to (3.5 Å−1, 3.1 Å) and (6.5 Å−1, 3.7 Å), respectively. When the pretreated sample was exposed to an NH3 or NO + NH3 atmosphere, the signal of the first shell sharply decreased (Fig. 3c, d), suggesting that the CNs of the Cu ions significantly declined due to their reduction. Moreover, the two lobes are not well-resolved in the spectra, indicating a decrease in the scattering from the second shell. This is consistent with the formation of dynamic [Cu(NH3)2]+ species, which is supported by the appearance of feature B in Supplementary Fig. 13a after NH3 or NO + NH3 adsorption. After oxidation by O2 and NO2, the CNs of the first shell increased to a level similar to that of the pretreated sample, accompanied by the formation of two well-resolved lobes at the second shell (Fig. 3e, f). This demonstrated that Cu(NH3)2+ species are oxidized into CuII ions and that the interaction between the Cu2+ ions and the zeolite framework is recovered. Besides the scattering by framework Si (or Al), the second lobe at 6.5 Å−1 probably resulted from the scattering of the second shell Cu species, since oxygen-bridged Cu dimers are formed after CuI(NH3)2 oxidation by O25,38. Compared with oxidation by O2, oxidation by NO2 resulted in a higher signal for the lobe at ~6.5 Å−1, indicating that more CuI species are oxidized into CuII ions (fw-CuII or NH3-solvated CuII species with high CNs) during the reaction with NO2. This phenomenon is consistent with the result reported by Paolucci et al. showing that NO2 can oxidize the residual CuI(NH3)2 species that cannot be oxidized by O2. As also reported by Paolucci, the transient oxidation of CuI(NH3)2 species by NO2 is a single-site process without formation of Cu dimers. Therefore, it can be inferred that the presence of NO2 probably changed the SCR reaction active sites from dimer Cu to isolated Cu species, which further influence the SSCR reaction. This deduction indicated that most Cu species are bonded with the zeolite framework and that the mobility of Cu species is limited during the process of CuI(NH3)2 oxidation by NO2. Although the transient reaction can reflect the Cu state and coordination during half-cycles, it was deemed more meaningful to identify the Cu species under FSCR reaction conditions.

Figures 4g–i depicts 2D plots of the WT EXAFS spectra under SSCR, FSCR and NO2-SCR conditions. Under SSCR conditions, the WT EXAFS spectra resemble the ones in Fig. 3c, d. The first shell peak weakened under SSCR conditions, indicating a decrease in the CNs of Cu species. The absence of the lobes at the second shell suggests the easy mobility of the copper complex due to the NH3 solvation effect. In the presence of NO2, however, the CNs of the first shell significantly increased, indicating the oxidation of copper species, which was also supported by the results of McEwen et al.26. Moreover, two well-resolved lobes at the second shell are observed, suggesting that oxidization leads to the copper species becoming closely coordinated with the zeolite framework, which limits their mobility during the SCR reaction. The WT EXAFS spectra are consistent with the Fourier-transformed (FT) EXAFS results (Supplementary Fig. 14 and Table 1), which are discussed in detail in the Supporting Information. The above results proved the existence of greater amounts of dynamic CuI(NH3)2 species under SSCR reaction conditions than that under FSCR and NO2-SCR reaction conditions. Notably, although we proved the existence of significant framework-bound CuII species under FSCR conditions, the NH3-solvated CuII species cannot be ruled out by the XAFS experiment. Indeed, the NH3-solvated CuII species existed, as indicated by the observation of NH3 desorption from Cu sites in FSCR-TPD profiles (Supplementary Fig. 5), which was consistent with the computed phase diagram reported by Paolucci et al.28. Therefore, we next turned to the DFT calculation to investigate the possible FSCR reaction pathways over fw- and NH3-solvated CuII [Cu2+ and (CuIIOH)+] species, BASs and dimer Cu species.

### DFT calculation

We first calculated the FSCR reaction pathway over fw-CuII species (Fig. 4). The framework-bound CuII first adsorbs two NH3 molecules without separation from the framework, which then interacts with NO2 to form Z2CuIINH3OH and NH2NO species (B → C). The Z2CuIINH3OH further adsorbs an NH3 molecule and reacts with NO, resulting in the formation of Z2CuIINH3, NH2NO and H2O (E → F), which was predicted to be the rate-determining step of the SCR reaction cycle with a high energy barrier of 1.92 eV. The formed NH2NO is easily decomposed into N2 and H2O through a series of H-migration and isomerization processes (Supplementary Fig. 16)29. Last, the gaseous NH3 molecules are supplied to regenerate the initial A species.

Next, the FSCR reaction pathway over ZCuIIOH was calculated and depicted in Fig. 5. ZCuIIOH first adsorbs an NH3 molecule to reach a coordinatively saturated state, which interacts with NO2 to form an HNO3 molecule without any energy barrier. The B species is actually considered to be NH4NO3 adsorbed on Cu sites. Next, the adsorbed HNO3 reacts with NO from the gas phase with an energy barrier of 0.87 eV, resulting in the formation of adsorbed HNO2 and the release of an NO2 molecule (C → D). Then, the adsorbed HNO2 reacts with the NH3 ligand to generate NH2NO and H2O. As the desorption of N2 and H2O molecules occurs, NH3 and NO2 are adsorbed at the Cu site and react to generate NH2NO and -OH groups. With the decomposition of NH2NO into N2 and H2O, the ZCuIIOH site is regenerated. The rate-determining step of the FSCR cycle over the ZCuIIOH site corresponds to the reaction of adsorbed HNO2 with an NH3 ligand to produce NH2NO and H2O (E → F), with an energy barrier of 1.58 eV.

In addition, the FSCR reaction pathways over NH3-solvated CuII species [CuII and (CuIIOH)+] were also calculated and presented in Supplementary Fig. 17, 18. All the energy barriers were found to be relatively high (1.54 and 1.65 eV). Moreover, we consider the possibility that various NH3-solvated CuII species diffuse into an adjacent cage to form CuII-pairs as shown in Supplementary Fig. S15. As expected, the formation of CuII pairs from CuII(NH3)4 and CuIINO2(NH3)3 is both thermodynamically and kinetically inhibited due to the steric effect as well as strong interaction with zeolite framework. However, it should be noted that CuIIOH(NH3)3 is different from the other forms since binuclear CuIIOH(NH3)3 in one cage is thermodynamically more stable than the isolated configuration. Villamaina et al. validated the formation of CuII(OH)(NH3)x dimeric complexes in the oxidation atmosphere through CO + NH3 titration experiment41. Hu et al. proposed that CuII(OH)(NH3), which is structurally similar to CuI(NH3)2+ that has one charge and two ligands, acts as inter-cage transportation medium6. The Z2CuII species can transform ZCuII(OH) by NH3-assisted hydrolysis to achieve the Cu pairing42. However, the regeneration of CuII(OH) in dimeric form showed a high energy barrier of 1.58 eV (A → B in Supplementary Fig. S19), suggesting that the dimeric CuII(OH) species were not highly active in the FSCR reaction.

The FSCR reaction pathway at BASs is displayed in Fig. 6. NH3 is adsorbed on the BASs to form NH4+ species. Two NO2 molecules interact with the NH4+ species to form NH4NO3 species and release an NO molecule without any energy barrier (A → B). The release of NO was also observed in our previous studies during NO2 adsorption on H-SSZ-13 zeolite43. Then, NO interacts with an NH3 from the gas phase to form an NH3 ∙ ∙ ∙ NO complex, which further reacts with NH4NO3 to form NH4+, HNO3, and NH2NO via an H-migration process (B → C). NH2NO is decomposed into N2 and H2O. Subsequently, HNO3 reacts with NO from the gas phase, resulting in the formation of HNO2 and the release of an NO2 molecule (E → F). Reaction between HNO2 and NH4+ species leads to the formation of an NH3 ∙ ∙ ∙ NO complex and an H2O molecule. The NH3 ∙ ∙ ∙ NO complex transfers an H atom to regenerate the BAS and changes into NH2NO, which then decomposes into N2 and H2O. The whole catalytic cycle is completed. The overall energy barrier of the FSCR over the Brønsted acid site is 1.27 eV, which corresponds to the reaction between NH4NO3 and the NH3 ∙ ∙ ∙ NO complex, much lower than that over various Cu sites. The DFT-calculated results indicate that the FSCR process in the SSZ-13 zeolite system tends to occur at BASs.

In summary, by combining the analysis of in situ spectroscopic measurements with DFT calculations, we found that NO2 leads to the deep oxidation of copper species as CuII species (fw-CuII and NH3-solvated CuII with high CNs), which significantly inhibits the mobility of Cu sites. As a result, the FSCR reaction occurs primarily at the BASs even though it has a higher energy barrier (1.27 eV) than the locally homogeneous SSCR reaction at dynamic sites (about 1.0 eV). This work reveals the origin of the abnormal NH3-SCR behavior over the commercial Cu-SSZ-13 catalyst in the presence of NO2.

## Methods

### Sample preparation

The initial Cu-SSZ-13 zeolite was in situ synthesized by a one-pot method15. The ratio of Na2O/Al2O3/H2O/SiO2/Cu-TEPA was 3.5/1.0/200/25/3 and the crystallization of the zeolite was performed at 120 °C for 5 days. Due to the excess Cu in the initial product, aftertreatments were required to optimize the Cu contents and distribution. In detail, the as-synthesized Cu-SSZ-13 was post-treated with 0.1 mol/L HNO3 at 80 °C for 12 h to remove CuOx species. After calcination at 600 °C, the sample was stirred in NH4NO3 solution (0.01~0.2 mol/L) at 40 °C for the second post-treatment process, followed by filtration, washing, drying and calcination at 600 °C. The obtained Cu-SSZ-13 catalysts were Al-rich zeolites with Si/Al of ~5 and various Cu loadings from 0.4 to 3.8 wt.% (Supplementary Table 2).

### Catalyst evaluation

The standard SCR (SSCR), fast SCR (FSCR) and slow SCR (NO2-SCR) reactions were carried out in a fixed-bed flow reactor system with an online Nicolet Is50 spectrometer, which was used to detect the concentrations of reactants and products. The SSCR conditions included 500 ppm NO and 500 ppm NH3; the FSCR conditions included 250 ppm NO, 250 ppm NO2 and 500 ppm NH3; the NO2-SCR conditions included 300 ppm NO2 and 500 ppm NH3. All the conditions included 3.5% H2O, 5%O2 and N2 balance. The total flow rate was 500 mL/min. The NOx (NO and NO2) conversion was calculated at steady state:

$${NO}\,{conversion}=\left(1-\frac{{\left[{NO}\right]}_{{out}}}{{\left[{NO}\right]}_{{in}}}\right)\times 100\%$$
(5)
$${{NO}}_{2}\,{conversion}=\left(1-\frac{{\left[{{NO}}_{2}\right]}_{{out}}}{{\left[{{NO}}_{2}\right]}_{{in}}}\right)\times 100\%$$
(6)
$${{NO}}_{x}\,{conversion}=\left(1-\frac{{\left[{{{{{\rm{NO}}}}}}\right]}_{{{{{{\rm{out}}}}}}}+{\left[{{{{{{\rm{NO}}}}}}}_{2}\right]}_{{{{{{\rm{out}}}}}}}}{{\left[{{{{{\rm{NO}}}}}}\right]}_{{{{{{\rm{in}}}}}}}+{\left[{{{{{{\rm{NO}}}}}}}_{2}\right]}_{{{{{{\rm{in}}}}}}}}\right)\times 100\%$$
(7)

To conduct the kinetic studies, the gas hourly space velocity (GHSV) was controlled by adjusting the catalyst weight. The GHSV of SSCR, FSCR and NO2-SCR were about ~800,000 h−1, ~1,000,0000 h−1 and ~2,000,000 h−1, respectively. The reaction rates (r) in this study were normalized by catalyst weight based on Eq. (8). The activation energies (Ea) were calculated by the Arrhenius Eq. (9).

$$r=\frac{{F}_{{{NO}}_{x}}\bullet {X}_{{{NO}}_{x}}}{{W}_{{cat}}}$$
(8)
$$r={\left[{{NO}}_{x}\right]}_{0}A{e}^{\left(-\frac{{Ea}}{{RT}}\right)}$$
(9)

where FNOx represents the NOx flow rate (mol/s), XNOx represents the NOx conversion, Wcat is the mass of the catalyst (g), and [NOx]0 is the inlet concentration of NOx. NOx represents NO, NO2 or a mixture of both.

### Characterization

The elemental composition of the catalysts was measured by inductively coupled plasma atomic emission spectroscopy (ICP-AES). N2 adsorption-desorption analysis of the samples was conducted on a Micromeritics ASAP 2020 instrument. The acid site distribution and contents were measured by NH3 temperature-programmed desorption (NH3-TPD) using the NH3-SCR activity measurement instrument described above. Samples of about 30 mg were used and pretreated in 10% O2/N2 at 500 °C for 30 min before cooling down to 120 °C. Then, the gas was changed to 500 ppm NH3/N2 for 60 min, followed by N2 purging for 60 min. Finally, the temperature was raised to 700 °C at a rate of 10 °C/min.

The in situ X-ray absorption fine structure (in situ XAFS) experiments were performed on the 1W1B beamline of Beijing Synchrotron Radiation Facility (BSRF). The absorption data from −200 eV to 800 eV of the Cu K-edge (8979 eV) were collected. The sample was first pretreated in O2/He at 500 °C for 30 min before decreasing the temperature to 200 °C, after which the Pre. spectra were collected. Then, the sample was exposed to 500 ppm NH3/He, 500 ppm NO/He and 500 ppm NH3/He + 500 ppm NO/He for 60 min, respectively, and spectra were collected. After reduction by (NO+NH3)/He, the sample was exposed to 5% O2/N2 and 500 ppm NO2/N2 for 60 min, respectively, to obtain the absorption data for the oxidized sample. Moreover, the in situ absorption data were collected after the pretreated samples were exposed to SSCR, FSCR and NO2-SCR atmospheres for 60 min. The X-ray absorption near-edge structure (XANES) data were background-corrected and normalized using the Athena module implemented in the IFFEFIT software package44. Extended X-ray absorption fine structure (EXAFS) data were analyzed and fitted using Athena and Artemis (3.0 < k < 13.0 Å−1). An amplitude reduction factor (S02) of 0.85 was used for all the fitted data sets. Wavelet transform (WT) analysis of the EXAFS was performed to precisely investigate the local coordination environment of copper species.

### Computational details

Spin-polarized periodic DFT calculations were carried out with the Vienna ab initio simulation package (VASP)45 The Perdew−Burke−Ernzerhof (PBE) generalized gradient approximation was adopted with the van der Waals correction proposed by Grimme (i.e., DFT-D3 method)46. The Kohn-Sham orbitals were expanded with a plane-wave basis set with a cutoff energy of 500 eV, and the plane augmented wave (PAW) method was used to describe the interaction between the valence electrons and the cores47. The DFT + U method was applied to Cu 3d states with Ueff = 6.0 eV to describe the on-site Coulomb interactions29,48. During geometrical optimization, the self-consistent-field electronic energies were converged to 1 × 10−5 eV and all other atoms were fully relaxed until the maximum force on the atoms was less than 2 × 10−2 eV/Å. The Brillouin zone was sampled with a Monkhorst-Pack k-point grid of 1 × 2 × 2. The Gaussian smearing method was utilized, with a smearing width of 0.2 eV. The transition states of elementary steps were located using the climbing image nudged elastic band (CI-NEB) method with several intermediate images between initial and final states49,50. Thermodynamic data were processed with the VASPKIT code51 and the Gibbs free energies were calculated at 200 °C. The SSZ-13 zeolite structure was modelled using two rhombohedral unit cells (24 tetrahedrally coordinated atoms) with size of 18.84 Å × 9.42 Å × 9.42 Å (Supplementary Fig. 20). One Si atom was replaced by one Al atom in each double 6-membered ring, resulting in a model with a Si/Al ratio of 11. One H atom was introduced onto one of the O atoms connected with each Al atom to keep the structure charge-neutral. Based on previous studies4,29, the present computational settings and models were reliable for investigating the NH3-SCR mechanism over Cu-SSZ-13 zeolites.