SO2 Poisoning of Cu-CHA deNOx Catalyst: The Most Vulnerable Cu Species Identified by X-ray Absorption Spectroscopy

Cu-exchanged chabazite zeolites (Cu-CHA) are effective catalysts for the NH3-assisted selective catalytic reduction of NO (NH3-SCR) for the abatement of NOx emission from diesel vehicles. However, the presence of a small amount of SO2 in diesel exhaust gases leads to a severe reduction in the low-temperature activity of these catalysts. To shed light on the nature of such deactivation, we characterized a Cu-CHA catalyst under well-defined exposures to SO2 using in situ X-ray absorption spectroscopy. By varying the pretreatment procedure prior to the SO2 exposure, we have selectively prepared CuI and CuII species with different ligations, which are relevant for the NH3-SCR reaction. The highest reactivity toward SO2 was observed for CuII species coordinated to both NH3 and extraframework oxygen, in particular for [CuII2(NH3)4O2]2+ complexes. Cu species without either ammonia or extraframework oxygen ligands were much less reactive, and the associated SO2 uptake was significantly lower. These results explain why SO2 mostly affects the low-temperature activity of Cu-CHA catalysts, since the dimeric complex [CuII2(NH3)4O2]2+ is a crucial intermediate in the low-temperature NH3-SCR catalytic cycle.

T he emission of nitrogen oxides (NO x ) from diesel vehicles is a global environmental challenge. 1,2 State of the art exhaust gas aftertreatment systems contain catalysts for selective catalytic reduction of NO x by ammonia (NH 3 -SCR), capable of reducing well over 90% of the NO x emitted by the engine. In the NH 3 -SCR reaction, NO reacts with NH 3 in the presence of O 2 to form N 2 and H 2 O. At present, Cu-exchanged chabazites (Cu-CHA) are the preferred catalysts for NH 3 -SCR, due to their superior low-temperature activity (150−350°C ) 3,4 and hydrothermal stability. 5,6 The temperature dependence of the NH 3 -SCR activity of Cu-CHA catalysts shows a minimum at around 350°C, which indicates that the reaction mechanism at low temperatures is different from that at higher temperatures. 7 The NH 3 -SCR reaction cycle for the low-temperature activity is a redox cycle, consisting of a series of oxidation and reduction steps, in which the oxidation state of Cu changes between Cu I and Cu II . The NO and NH 3 coordinate to Cu in the zeolite, giving rise to a variety of Cu species along the NH 3 -SCR cycle. 8−11 The low-temperature activity of Cu-CHA catalysts originates from the ability to form mobile Cu I (NH 3 ) 2 complexes under SCR conditions. Pairs of these species constitute the active Cu sites capable of O 2 activation via the formation of [Cu II 2 (NH 3 ) 4 O 2 ] 2+ dimers around 200°C, which is a crucial step in the NH 3 -SCR reaction cycle. 12,13 In practice, the application of Cu-CHA catalysts for the NH 3 -SCR is restricted to ultralow-sulfur diesel fuels, due to the fact that a few ppm of SO 2 present in the exhaust gas drastically reduces the activity at low temperatures. 3,4,14 Multiple studies show that SO 2 affects the Cu mobility, the amount of Cu active sites, 14 and the redox behavior of the Cu in the NH 3 -SCR cycle. 9,15 Most studies have focused on the overall effect of SO 2 on the performance of the catalysts, 14−21 while the chemistry behind SO 2 poisoning at the molecular level remains poorly understood. To determine a mechanism for SO 2 poisoning, one must identify the species in the Cu-CHA catalysts that interact with SO 2 . To this end, we have selectively prepared well-defined Cu I and Cu II species with different ligands inside the pores of the Cu-CHA catalyst and exposed them to SO 2 under well-defined conditions. We monitored the changes in the Cu K-edge X-ray absorption spectra (XAS) during the absorption of SO 2 . This allowed us to determine the chemical state of the Cu that interacts with SO 2 . The results were corroborated by X-ray emission spectroscopy (XES) and measurements of the SO 2 uptake using temperature-programmed desorption (TPD) of SO 2 .
The Cu-CHA catalyst used in this study had a Si/Al ratio of 6.7 and a Cu loading of 3.2 wt % (Cu/Al = 0.24). The Cu Kedge XAS and Cu Kβ valence-to-core XES measurements were carried out at the BM23 22 and ID26 23 beamlines of the European Synchrotron Radiation Facility (ESRF), respectively. Sample treatment protocols consisted of three distinct steps. First, all samples were heated to 550°C in a 10% O 2 /He flow, removing water and forming Cu II species bound to the framework of the zeolite (fw-Cu II ). Then, the specific state of Cu was prepared, using one of the six different pretreatment procedures summarized in Table 1. Finally, the catalyst was exposed to 400 ppm SO 2 /He flow at 200°C for 3 h until no visible changes in the spectra occurred. Further experimental details are given in the Supporting Information.
The Cu species formed with the pretreatments differ in three aspects: (1) the oxidation state of Cu (Cu I or Cu II ), (2) the coordination of the Cu (NH 3 or/and O), and (3) the interaction of the Cu with the framework (fw-coordinated or mobile species). Figure 1 shows the evolution of Cu K-edge XANES and EXAFS spectra during the exposure of the pretreated Cu-CHA catalyst to 400 ppm SO 2 /He flow at 200°C. For all Cu I species and fw-Cu II species (procedures 1−4 in Table 1), only minor changes are observed upon SO 2 exposure, indicating that these species are not very reactive toward SO 2 . In contrast, for Cu II species in the presence of NH 3 (procedures 5 and 6) significant changes are observed in the spectra. In these cases, the exposure to SO 2 results in a pronounced increase of the XANES peak at 8983 eV, characteristic for linear Cu I complexes, 9,26,27 and a decrease in the intensity of the first shell in the EXAFS FT. This means that some of the Cu II species are reduced to Cu I upon interaction with SO 2 . The decrease in the first-shell intensity indicates a reduction of the coordination number for the Cu ions, which is also in line with the formation of a linear Cu I species.
The species obtained in procedure 5 are the oxygen-bridged diamine dicopper complexes [Cu II 2 (NH 3 ) 4 O 2 ] 2+ , which are formed by the reaction of O 2 with a pair of [Cu I (NH 3 ) 2 ] + complexes. 13 In the reaction cycle for the low-temperature NH 3 -SCR reaction, 11 (Figure 2). The necessary stock of available oxygen needed for the formation of the mixed-ligand species is expected to be present in the sample, as a wavelet analysis of the EXAFS collected after heating to 550°C and cooling to 200°C in 10% O 2 /He flow reveals the presence of Cu−Cu scattering usually attributed to the oxygen-containing dimers 29,30 ( Figure S8 in the Supporting Information), which may be susceptible to form mixed-ligand species upon exposure to NH 3 .
The evolution of XANES spectra upon interaction with SO 2 shows that the most susceptible species are Cu II with mixed (NH 3 ) x O y ligation, whereas Cu I species or Cu II in the absence of NH 3 are much less affected. These findings are supported by X-ray adsorbate quantification (XAQ) data, 31 collected simultaneously with the XAS measurements during the exposure to SO 2 , and a TPD analysis of a parallel set of catalyst samples, exposed to the same pretreatments used in XANES experiments (Figure 3a). We find the highest sulfur content (S/Cu ratio) for the [Cu II 2 (NH 3 ) 4 O 2 ] 2+ and Cu II + NH 3 procedures. The sulfur uptake of the [Cu I (NH 3 ) 2 ] + and fw-Cu II moieties was ca. 3 times lower, and for the bare fw-Cu I species, it was ca. 6 times lower. These results show that the reaction between the [Cu II 2 (NH 3 ) 4 O 2 ] 2+ species and SO 2 contributes the most to the accumulation of SO 2 in the Cu-CHA catalyst.
Interestingly, the sulfur content in the Cu II + NH 3 sample lies between those for samples with pure [Cu I (NH 3 ) 2 ] + and [Cu II 2 (NH 3 ) 4 O 2 ] 2+ species, which in combination with the linear combination fit shown in Figure 2 suggests that the reactivity of the Cu II (NH 3 ) x O y species obtained after Cu II + NH 3 treatment toward SO 2 is similar to that of [Cu II 2 (NH 3 ) 4 O 2 ] 2+ . By comparing the SO 2 -TPD curves of Cu-CHA samples with that of (NH 4 ) 2 SO 4 , adsorbed on Cu-free CHA ( Figure  3b), we can also deduce that the elevated sulfur content in the samples with the Cu II (NH 3 ) x O y species is due to the reactivity toward SO 2 and not to the formation of (NH 4 ) 2 SO 4 in a reaction of SO 2 with NH 3 and NH 4 + groups stored in the zeolite framework. For the adsorbed (NH 4 ) 2 SO 4 , we observe SO 2 desorption at around 380, 530, and 1000°C (gray curve in Figure 3b). The desorption at 380°C matches the known thermal decomposition of (NH 4 ) 2 SO 4 ; 32 the other two peaks are probably due to the interaction of either (NH 4 ) 2 SO 4 or products of its decomposition with the zeolite, their precise interpretation being beyond the scope of the present argument. For all three Cu-CHA samples containing NH 3 before exposure to SO 2 ([Cu I (NH 3 ) 2 ] + , [Cu II 2 (NH 3 ) 4 O 2 ] 2+ , and (Cu II + NH 3 ) procedures), we observe SO 2 desorption at around 420°C ( Figure 3b). As this does not match any of the observed desorption characteristics of (NH 4 ) 2 SO 4 in Cu-free Cu-CHA, the SO 2 -TPD feature at 420°C reflects an interaction of Cu with SO 2 . Interestingly, the SO 2 -TPD curve for the sample with the dominant fw-Cu II species shows a significant SO 2 desorption peak close to 1000°C, which, together with the lack of changes in Cu K-edge XANES upon exposure to SO 2 , indicates the formation of some sulfur deposits not directly coordinated to Cu.
The presence of Cu−N and Cu−O bonds in the [Cu II 2 (NH 3 ) 4 O 2 ] 2+ complex has been independently confirmed by valence-to-core XES. 27,33,34 XES spectra at different stages of pretreatment leading to the formation of [Cu II 2 (NH 3 ) 4 O 2 ] 2+ dimers are reported in Figure 4. The origin of the Kβ′′ satellite peak is the transition from the ligand s orbitals to Cu 1s, which makes its position sensitive to the species directly coordinated to Cu and allows it to discriminate   Figure 4 shows that after heating in O 2 Cu is predominantly coordinated by oxygens (as expected for the fw-Cu II species), whereas after exposure to NO + NH 3 N ligands are dominating, as expected for a [Cu I (NH 3 ) 2 ] + linear complex. After subsequent exposure to O 2 and formation of [Cu II 2 (NH 3 ) 4 O 2 ] 2+ dimers, the peak broadens, confirming the presence of both Cu−N and Cu−O bonds. These bonds remain after exposure to SO 2 , while no significant contribution from Cu−S bonds 38 is observed, suggesting that the possible SO 2 binding to the Cu is carried out through an oxygen atom.
In conclusion, the in situ XAS and XES measurements of different Cu intermediates formed in a Cu-CHA catalyst exposed to SO 2 demonstrate that Cu II species with mixed NH 3 and O ligation of Cu are particularly reactive toward SO 2 , whereas Cu I species and Cu II without NH 3 are much less affected by it. In particular, the [Cu II 2 (NH 3 ) 4 O 2 ] 2+ complex, which is formed upon activation of O 2 in the NH 3 -SCR cycle, shows a clear reaction with SO 2 , resulting in a partial reduction of the Cu II and accumulation of sulfur in the zeolite. Therefore, we conclude that this reaction is responsible for the poisoning of Cu-CHA catalysts in NH 3 -SCR by SO 2 .
Experimental details, XANES linear combination fits before SO 2 exposure, wavelet transform analysis of the sample heated in O 2 , and EXAFS fitting results (PDF)