Operando Spectroscopic Studies of Cu–SSZ-13 for NH₃–SCR deNOx Investigates the Role of NH₃ in Observed Cu(II) Reduction at High NO Conversions

SSZ-13 zeolite (Si/Al = 13) was synthesized as described previously, using N,N,N-trimethyladamantammonium hydroxide as a structural directing agent in a fluoride media, under static hydrothermal conditions [13, 21, 23]. The proton form of the zeolite is obtained by calcining the sample in air by heating at 1 °C min⁻¹ to 120 °C, held for 2.5 h and then at 4 °C min⁻¹ to 550 °C, held for 10 h. The copper exchanged form was prepared using a wet ion exchange method in which typically an amount of the H-SSZ-13 would be added to an aqueous solution of copper sulphate (50 ml of a 0.1 M solution of CuSO₄ per gram of zeolite) and heated at 80 °C for 2 h whilst stirring. The product is then recovered by vacuum filtration, washed with copious amounts of water, dried overnight at 80 °C and calcined in air using the procedure outlined above (see ESI for further details). Cu–SSZ-13-a (2.92 wt% Cu; Si/Al = 13) is a sample which was prepared by following this procedure once and the material is 75% Cu ion exchanged in to available H⁺ sites. Cu–SSZ-13-b (3.86 wt% Cu: Si/Al = 12) was prepared by sequentially performing 4 ion exchange and calcinations on a sample obtained from Johnson Matthey UK, and is a material in which 100% Cu ion exchanged in to available H⁺ sites has occurred. The materials consist of well-defined rhombohedral crystals with good phase purity and crystallinity and contain isolated Cu(II) species upon dehydration, with no evidence of particles of copper or copper oxide present in the UV–vis spectra (For further information about synthesis, characterisation and catalytic testing see ESI). Prior to performing operando spectroscopic analysis, the zeolites were pressed in to pellets (8 mm bore, 1.5 tonne pressure), broken up and sieved, retaining the 250–440 µm fraction for experiments.

Cu K-edge XAS studies were performed at the Diamond Light Source (DLS), UK, [24] on beamline B18 and at the Swiss Light Source (SLS) at the Paul Scherrer Institute, Switzerland, on the SuperXAS beamline [25]. At the DLS, measurements were performed in transmission mode using ion chamber detectors with a fast scanning (QEXAFS) Si(111) double crystal monochromator. Each spectrum took 60 s to acquire with a Cu foil placed between It and Iref. At the SLS the Si(111) crystal of the quickXAS monochromator was oscillated at 5 Hz and the spectra averaged over 1 min. At both beamlines, the sieved zeolite was placed in to a quartz capillary based gas flow cell (3 mm diameter at diamond, 0.8 mm diameter at SLS), the zeolite material was placed in the reactor cell between two quartz wool plugs to prevent physical movement of the material under flowing gas conditions. The capillary was sealed in to the gas flow cell by screw fittings to ensure a gas tight connection and a thermocouple was inserted into the end of the cell such that it reached the centre point of the catalyst bed. The cell was mounted into the beamline so that the beam was focussed towards the front section of the catalyst bed, while ensuring the beam was fully focussed within the sample (beam size 0.5 × 0.5 mm at SLS, 0.05 × 0.05 mm at DLS), temperature control was achieved using hot air blowers and attached to heated gas lines controlled by MFCs, the outlet of the gas cell was connected to a mass spectrometer (MKS mass spectrometer at DLS, Hiden at the SLS) that was used to follow signals of m/z 2 (H₂), 4 (He), 18 (H₂O), 28 (CO, N₂), 30 (NO), 32 (O₂), 44 (N₂O/CO₂), and 46 (NO₂).

Prior to the operando experiments, the calcined catalysts were subjected to high temperature activation in an O₂ atmosphere. Samples were heated to 400 °C at 10 °C min⁻¹ and held at this temperature until no more changes were observed in the XAS spectra, the sample was then allowed to attain the desired temperature. At DLS a temperature ramp experiment was conducted during which the sample was heated from 150 to 600 °C at 5 °C min⁻¹, during this temperature ramp the sample was exposed to an atmosphere of NH₃ (3 ml min⁻¹, 5% in He), NO (15 ml min⁻¹, 1% in He), O₂ (28.571 ml min⁻¹, 17.5% in He) and He (3.428 ml min⁻¹) giving a total flow of 50 ml min⁻¹, 3000 ppm of NH₃ and NO in 10% O₂ over a packed bed, of 8 mm in length with a radius of 1.5 mm, giving a GHSV of 53,000 h⁻¹. NO conversion was calculated as a function of intensity of NO mass spec signal and the intensity of equilibrated NO mass spec signal at low temperature \(\left[ {\left( {\left( {{{\text{I}}_{{\text{ref}}}} - {{\text{I}}_{{\text{temp}}}}} \right){\text{/}}{{\text{I}}_{{\text{ref}}}}} \right)\times100\%} \right].\) At the SLS experiments based on varying the GHSV, at a constant temperature of 250 °C, were conducted under conditions for both NH₃–SCR and NH₃ + O₂. During the NH₃–SCR experiments the zeolite catalysts were exposed to the following gas conditions, NH₃ (2.5 ml min⁻¹, 1% in N₂), NO (2.5 ml min⁻¹, 1% in N₂), O₂ (2 ml min⁻¹, 100%) and N₂ (12.8 ml min⁻¹) giving a total flow of 19.8 ml min⁻¹, 1262 ppm of NH₃ and NO in 10% O₂ over a packed bed of 7 mg of zeolite (density of zeolite = 0.526 g cm⁻³_, volume of bed = 0.0133 cm⁻³), giving a GHSV of 90,000 h⁻¹. Under a second experiment at higher GHSV the samples were exposed to NH₃ (6.5 ml min⁻¹, 1% in N₂), NO (6.5 ml min⁻¹, 1% in N₂), O₂ (6 ml min⁻¹, 100%) and N₂ (32 ml min⁻¹) giving a total flow of 50 ml min⁻¹, 1262 ppm of NH₃ and NO in 10% O₂ over a packed bed of 7 mg of zeolite, giving a GHSV of 225,000 h⁻¹. The two GHSV experiments were repeated under conditions in which both NH₃ and O₂ were present, in which the gas composition remained the same with the removal of the NO and addition of extra N₂ to ensure the same GHSV were maintained (full details of the gas flow calculations can be found in the ESI). EXAFS data processing was performed using IFEFFIT [26] with the Horae package [27] (Athena and Artemis).

A temperature ramp experiment was conducted on an activated sample of Cu–SSZ-13-a during NH₃–SCR of NO. Initially the sample was held at 150 °C and exposed to a constant stream of gas that comprised; 3000 ppm of NH₃ and NO in 10% O₂ at a GHSV of 53,000 h⁻¹. The temperature was then increased at 5 °C min⁻¹ to 600 °C. The results obtained are shown in Fig. 2. The operando XANES results show that after activation and prior to catalyst activity the spectra contain a small pre-edge shoulder between 8985 and 8990 eV assigned to the Cu²⁺ 1s/4p transition (see Fig. 3a), which is evidence of a move of the Cu²⁺ to a lower symmetry position in the zeolite framework and is indicative of total dehydration and the Cu ions adopting specific crystallographic positions. The temperature at which the catalyst starts to become active (around 200 °C) corresponds to a rapid change in the XANES spectra as a second large pre-edge peak appears with an intense sharp feature at 8982 eV. This feature has been reported to be due to a transition from a 1s to the doubly degenerate 4_pxy orbitals in a two-coordinate Cu⁺ system [28‐31]. It has been suggested that this feature could explain increased deNOx activity at 200 °C via enabling a fast SCR mechanism [30, 32]. The XANES data presented herein shows that there is a strong temperature dependency regarding the feature at 8982 eV and that heating Cu–SSZ-13-a under standard SCR reaction conditions above 225 °C results in the loss of the Cu⁺ leaving only Cu²⁺ species present. As this temperature corresponds with maximum NO conversion it would suggest that Cu²⁺ is the predominant species present during the standard NH₃–SCR of NO above temperatures > 200 °C.

Although the feature at 8982 eV rapidly decreases in intensity, it is observed that in the region of maximum catalytic conversion of NO, between 250 and 350 °C there is a small but distinguishable pre-edge peak still present (see Fig. 3a). Although the nature of the Cu(I) has been studied at low NO conversion under standard SCR conditions [33] the cause of the Cu(I) in this temperature regime, which corresponds to maximum NO conversion, has not been adequately ascribed to the mechanism of NH₃–SCR. To probe this feature further additional isothermal operando experiments were performed at different GHSV, of approximately 100,000 and 225,000 h⁻¹ at 250 °C (see Fig. 3b). Curiously in the additional experiments the feature at 8982 eV is not present at 100,000 h⁻¹ after 10 min but does start to appear at 20 min exposure. At 225,000 h⁻¹ the feature is present after 10 min then increases in concentration by 20 min of exposure however by 70 min exposure this apparent intensity has decreased.

The observed result, of partial Cu(II) reduction to Cu(I), during the initial exposure of the material to reagents, coupled with a change in intensity of the Cu(II) reduction at different GHSV implies that there is a change in the equilibrium of the composition of reactants present in the catalyst, and that at this leads to an environment which can reduce some of the Cu(II) present. The compositional change with in the zeolite material influences the catalytic performance of the material in terms of the conversion of NO. Increasing the GHSV of the gas stream through the catalyst decreases the conversion of NO modestly from 97% at 53,000 h⁻¹ to 83% at 225,000 h⁻¹ (see ESI for further details), the results are summarised in Table 1. The observed results in the 225,000 h⁻¹ experiment, in which the Cu(I) feature intensifies between 10 and 20 min and then decreases dramatically by 70 min exposure, suggests that the material is not under true steady state conditions for a significant amount of time after initial exposure. During this initial period, it can be suggested that NH₃ is being preferentially adsorbed (due to much stronger adsorbent adsorbate interactions than NO or O₂) and therefore during this initial period the concentration of adsorbed NH₃ is much higher than the other components present creating an environment in which Cu(II) can be reduced. It should be noted that the apparent decrease in the pre edge feature intensity between the 100,000 h⁻¹ experiment reported here and the 53,000 h⁻¹ experiment reported earlier may be due to the different gas compositions used for the experiments at DLS versus the SLS, this was due to the limitations of the minimum flow available to the MFCs at DLS coupled with the gas concentrations meant the minimum quantity of NH₃ and NO that could be used was 3000 ppm whereas at the SLS the experiments were operated at approximately 1000 ppm of NO and NH₃ which is closer to conditions present in exhaust streams of lean burn diesel vehicles [22].

If it is assumed that the stable Cu(I) is independent of the standard SCR reaction, its origin could potentially arise from either the interaction with one of the species present in the gas mixture or through another side reaction for example NO oxidation or NH₃ oxidation. It has been shown that in Cu–SSZ-13 for NO oxidation to occur it is necessary for the Cu species to be close enough in space to form dimeric species ([Cu–O–Cu]²⁺), [34] at the loadings reported herein (2.92–3.6%) the Cu speciation is below the required Cu loading for NO oxidation to be occurring in a significant amount [35]. Since a significant reduction of the Cu(II) too stable Cu(I) species is observed in the XANES data it was postulated that this may be due to the interaction of the NH₃ (in the presence of O₂) and the Cu(II) sites. Therefore, in situ XANES experiments were conducted at appropriate conditions in which NH₃ and O₂ were simultaneously present in the gas stream, to probe if the presence of Cu(I) was related to processes involving NH₃, including adsorption, oxidation or decomposition (see ESI for further details regarding NH₃ oxidation). It should be noted that the competitive non selective NH₃ oxidation is usually observed at temperatures significantly higher than 250 °C [36]. Results of the experiments are shown in Fig. 4, in which it is observed that Cu(I) is present at both GHSVs and the intensity is increasing with both increasing GHSV and time on stream.

It has been proposed recently that NH₃ consumption is a potential route to stable linear Cu(I) species via the following reaction [37]:

This would explain the observed features in the XANES data, however, if it were possible for this to be occurring at all Cu species present in the zeolite it should be anticipated that the intensity of the Cu(I) peak present in the XANES spectra would be much greater than is observed (i.e. if all Cu(II) were reduced). A possible explanation for this could be that it has previously been shown that Cu–SSZ-13-a has two different Cu sites occupied in the crystal structure at 250 °C one in the six ring and one in the centre of the eight rings present in chabazite and it may be possible for the NH₃ activated Cu reduction to be occurring at only one of these sites [18].

To test this hypothesis Cu–SSZ-13-b was prepared, in which the material has been ion exchanged a further three times and has a lower thermal barrier to Cu migration meaning that at 250 °C all the Cu ions present are in the six ring sites. Cu–SSZ-13-b was tested under conditions for both NH₃–SCR and NH₃ + O₂, the results of these experiments are shown in Fig. 5. Under SCR conditions there is no evidence of Cu(I) species present in the Cu–SSZ-13-b while under the simultaneous exposure of NH₃ with O₂ there is only a very small feature centred at 8982 eV present which is far less pronounced than observed under similar conditions for the Cu–SSZ-13-a. These results would suggest that the Cu reduction observed under certain reaction conditions is due to Cu ions present in the eight ring of the chabazite structure, and that by extension, Cu ions present in the six rings cannot undergo NH₃ mediated reduction. While these results do not rule out the possibility of Cu ions in the eight ring being able to participate in the NH₃–SCR of NOx the ability for the Cu in the eight ring sites to be reduced through an interaction with NH₃ may hinder the performance of the 8-ring site relative to the six rings sites in the NH₃–SCR of NOx.