Appraising Multinuclear Cu2+ Structure Formation in Cu-CHA SCR Catalysts via Low-T Dry CO Oxidation with Modulated NH3 Solvation.

Cu2+ ions (ZCu2+ (OH)- , Z2 Cu2+ ) are regarded as the NH3 -SCR (SCR=selective catalytic reduction) active site precursors of Cu-exchanged chabazite (CHA) which is among the best available catalysts for the abatement of NOx from Diesel engines. During SCR operation, copper sites undergo reduction (Reduction half-cycle, RHC: Cu2+ →Cu+ ) and oxidation (Oxidaton half-cycle, OHC: Cu+ →Cu2+ ) semi cycles, whose associated mechanisms are still debated. We recently proposed CO oxidation to CO2 as an effective method to probe the formation of multinuclear Cu2+ species as the initial low-T RHC step. NH3 pre-adsorption determined a net positive effect on the CO2 production: by solvating ZCu2+ (OH)- ions, ammonia enhances their mobility, favoring their coupling to form binuclear complexes which can catalyze the reaction. In this work, dry CO oxidation experiments, preceded by modulated NH3 feed phases, clearly showed that CO2 production enhancements are correlated with the extent of Cu2+ ion solvation by NH3 . Analogies with the SCR-RHC phase are evidenced: the NH3 -Cu2+ presence ensures the characteristic dynamics associated with a second order kinetic dependence on the oxidized Cu2+ fraction. These findings provide novel information on the NH3 role in the low-T SCR redox mechanism and on the nature of the related active catalyst sites.

Introduction

Air pollution is a growing worldwide concern, and abatement technologies are required to cope with increasingly stringent emission regulations. Nitrogen oxides (NOx) are a family of gaseous polluting compounds originating from combustion processes, and the NH3/urea selective catalytic reduction (SCR) is the leading, state‐of‐the‐art technology for their control from vehicular lean Diesel engines.[ , ] Cu2+ ions are considered the deNOx active site precursors in Cu‐exchanged chabazite‐type zeolites (Cu‐CHA), currently among the best performing catalysts for this application due to their superior hydrothermal stability, resistance to hydrocarbon poisoning and an effective SCR activity in a wide temperature range.[ , , ] In the last decade, several studies have been conducted to disclose the fundamental and mechanistic features of the SCR chemistry. However, many open issues still remain, in particular related to the nature of the Cu intermediates.[ , , ] Several works converged to highlight the SCR process redox nature, which relies on two half cycles concurrently proceeding during common working conditions: a reduction and an oxidation half cycle (RHC and OHC), where the ions are reduced to Cu+ and re‐oxidized back to Cu2+, respectively.[ , , , , ] The existence of two kinds of isolated Cu2+ ions located in different zeolite structure positions has been probed by a combination of transient methods, spectroscopic techniques and molecular dynamics,[ , , ] namely ZCu2+(OH)−, located in the eight‐membered ring of the CHA cage, where the two positive charges of copper are balanced by one coordination with the local framework‐Al (Z) and by a hydroxyl group (OH)−; and conversely, Z2Cu2+, which is bonded twice to the zeolite in the six‐membered ring. Furthermore, the relative populations of these two species are affected by several factors, such as hydrothermal aging as well as the SiO2/Al2O3 and the Cu/Al ratio of Cu‐CHA.[ , , ] In the presence of NH3, the ions become solvated and detached from the framework, forming Cu2+(OH)−(NH3)x mobile complexes. Recently, we proposed the dry CO oxidation protocol as a viable technique to explore the formation of dinuclear Cu2+ structures, originating from ZCu2+(OH)−, under conditions representative of the low‐T RHC regime over Cu‐CHA. As the oxidation requires a two‐electron exchange, whereas the reduction of each isolated Cu2+ involves one‐electron transfer, CO2 formation was suggested to signal the active participation of Cu2+ dimers in Cu‐zeolite catalysts.[ , , ] Furthermore, NH3 adsorption prior to the CO feed greatly enhanced the CO oxidation activity. This positive effect was attributed to the solvated ions’ mobility, as the ions were proposed to diffuse inter the cages, forming binuclear species capable to catalyze the CO oxidation. The dual site nature of the Cu2+ reduction process results are coherent with the recent RHC transient kinetic analysis over several of Cu‐CHA catalysts, which revealed a rate quadratic dependency of Cu reduction on the oxidized Cu2+ fraction once NH3 was pre‐loaded. Conversely, these findings do not reconcile with the widely proposed single‐site‐RHC mechanisms, suggesting the need for proposing novel, alternative pathways.[ , , ]

Herein, we have further inspected the role of NH3 in the formation of multinuclear Cu2+ species by adopting a combination of transient response methods. Dedicated tests with modulated NH3 coverage enabled loading the different catalyst adsorption sites to varying degrees. Subsequent dry CO oxidation tests clearly titrated the active dimer‐like Cu2+ species onset only in the presence of NH3‐ligated Cu2+, whereas NH3 adsorbed onto Brønsted sites did not significantly affect the CO oxidation process. The adopted protocols also provide information on the relative rates of NH3 adsorption, Cu‐dimer formation, and CO oxidation.

Results and Discussion

The goal of this research activity is to assess how the formation of multinuclear Cu2+ species is affected by the presence of NH3 coordinated to different acidic catalytic sites of Cu‐CHA, under conditions representative of low‐T SCR‐RHC. A deeper understanding of the NH3 adsorption mechanism on the Cu‐CHA sites is thus require, hence a series of dedicated tests have been carried out first. The details of these have been reported in the Experimental Section of this paper.

NH3 adsorption – Variable adsorption time effect

NH3, the primary reducing agent for NOx in NH3‐SCR, is also extensively used in literature to titrate the acid sites of the Cu‐CHA catalyst, exploiting its capacity to form chemical bonds with different strengths.[ , ] Thus, isothermal NH3 adsorption+TPD (temperature‐programmed desorption) tests allow to estimate both the total NH3 storage and the catalyst sites involved in the adsorption process. All the tests have been herein realized under dry conditions to avoid competition of H2O and NH3 for adsorption on the catalyst sites, Cu2+ hydrolysis phenomena and other complexities.[ , ] The isothermal adsorption and desorption phases were realized in the presence of O2 to avoid any Cu2+ reduction by NH3. Since the reference CO oxidation test is performed at isothermal conditions (200 °C), the same temperature has been adopted in these NH3 adsorption tests for consistency. An initial test has been realized with full catalyst saturation: the temporal evolution of the temperature and the involved species is shown in Figure S2 (Supporting Information), while Table S1 reports the quantitative analysis generated by NH3 and N2 profiles integration. During the initial NH3 feed, a dead time was observed (≈9 min), where all the reactant fed to the catalyst is adsorbed and no release is detected by the instruments. After this initial phase, the NH3 outlet concentration rapidly increased, reaching the feed concentration level at complete catalyst saturation. The cardinal phase for this work is the TPD step, which allows to evaluate the features of chemically adsorbed NH3. It is thus realized after an isothermal desorption phase to remove the contribution of weakly adsorbed species. To avoid ammonia oxidation during the temperature ramp, the O2 feed is cut off prior to the catalyst heating. A dual‐site desorption trend is identified during the TPD, a known trait for this materials already observed in previous experimental campaigns on a variety of Cu‐CHA samples with different SAR and Cu loadings.[ , ] The high‐temperature peak (centred at ≈450 °C) is associated with the NH3 desorption from Brønsted acid sites (ZH+), characterizing the unexchanged CHA framework, while the low‐temperature feature (≈350 °C) is ascribed to the Cu2+ ions’ Lewis acidity (with a possible minor contribution of extra‐framework Al). Concurrently with the NH3 chemical desorption, a N2 release is observed in the Lewis peak region. The integral nitrogen production is the result of full Cu2+ reduction by NH3, according to Cu2+ : N2=1 : 6 stoichiometries for both ions as reported in the following (Equations (1) and (2)) (Cu2+/N2=4.4 μmol/0.67 μmol=6.6:[ , ]

Accounting for this reduction process contribution, the overall integral NH3 TPD is evaluated as the sum of the NH3 release with twice the N2 production (10.4 μmol). As expected, this value is lower (−36 %) than the same assessed in a previous investigation where the adsorption was effected at 150 °C.[ , ] In an earlier work, we have emphasized the correlation between the Lewis desorption feature and the ammonia coordination of the copper ions, performing an analysis of the NH3/Cu ratio. This is typically used as a Cu2+ speciation index for Cu‐CHA catalysts; however, depending on the gas phase conditions, different numbers of NH3 ligands can be present on the Cu2+ ions. Lewis‐NH3/Cu shows a range between 3–4 when NH3 is present in the gas phase (forming Cu2+(OH)−(NH3)3 and Cu2+(NH3)4), while it decreases to 1–2 when gaseous ammonia is removed (Cu2+(OH)−(NH3), Cu2+(NH3)2).[ , ] Recently, we showed that the Lewis‐NH3/Cu ratio of different Cu‐CHA catalysts, assessed by deconvolution of the TPD curves, was in agreement with the expected number of ligands on ZCu2+(OH)− and Z2Cu2+. With this knowledge, we replicated the NH3 adsorption tests, changing the NH3 injection phase duration to evaluate the impact of the adsorption sites fractional coverage on the TPD phase. In the previous test with complete catalyst saturation, NH3 breakthrough occurred after an initial 9 min dead time. Thus, we chose reduced duration feed times of 1, 2, 4, 6 and 7 min. The respective isothermal NH3 adsorption and desorption phases are depicted in Figure 1A. Up to 6 min, no desorption was detected, with a minor release occurring at 7 min. Consequently, all the NH3 fed is directly stored on the catalyst. The corresponding TPD traces are plotted in Figure 1B as NH3+2*N2, alongside the TPD trace collected after full saturation. A single high‐temperature peak is detected considering 1 and 2 min of adsorption time, while only at 4 min the Lewis NH3 feature starts to appear. The Brønsted acid sites reach nearly complete saturation at 6 min, while the low‐temperature acidity proceeds to continuously increase. Accounting for the global NH3 adsorption mechanism on Cu‐CHA materials, this data set clearly points out how the Brønsted acid sites are saturated first, while NH3 coordinates to the Lewis sites only later on, as the weak adsorption sites are expected to be saturated last. Table 1 reports the quantitative analysis of the TPD traces for different adsorption times. Considering the trace after full NH3 saturation as reference, the NH3 loading (θNH3) is defined as the ratio between the TPD at a specific adsorption time with respect to full saturation conditions. On comparing Table 1 with Figure 1, it is possible to conclude that the solvation of the Cu2+ sites only initiates at NH3 loadings higher than 50 %, while, at lower θNH3 values, their contribution to the NH3 storage can be considered as negligible.

Figure 1

Comparison of NH3 adsorption/desorption phases with variable adsorption time (A), and the relative NH3+2*N2 TPD (B). NH3=500–0 ppm, O2=8–0 % v/v, H2O=0 % v/v, balanced in He. GHSV=266250 cm3/h/gcat (STP) (Wcat=16 mg). T=190–600 °C, heating rate=15°/min. Pre‐oxidized catalyst.

Table 1

Quantitative analysis of the TPD phase, with the evaluation of NH3 coverage (θNH3) for each adsorption time.

Variable NH3 ads. time [min]	TPD : NH3+2*N2 [μmol]	θNH3 [%]
Saturation	10.4	100
7	9.7	93
6	9.0	86
4	5.6	54
2	2.4	23
1	1.0	10

CO oxidation – Variable NH3 loading effect

The dry CO oxidation protocol has been already applied to the sample herein employed, and the temporal evolution of the global species has been reported elsewhere. [17] A net positive effect on CO2 formation was observed when the catalyst was fully saturated with NH3 prior to the CO feed. The same tests, with and without NH3, have been replicated: when CO reaches the catalyst, a transient CO2 production is detected (shown in Figure 2) with a characteristic initial peak and a slow decreasing trend. The presence of NH3 (θNH3=100 %) clearly enhances the process with respect to its absence (θNH3=0 %), in agreement with previous findings.

Figure 2

CO2 released during the CO oxidation phase with different NH3 loadings. Feed: CO=1000 ppm, O2=0 % v/v, H2O=0 % v/v, balanced in He. GHSV=266250 cm3/h/gcat (STP) (Wcat=16 mg). T=200 °C. Pre‐oxidized catalyst.

As anticipated in the Introduction section, the NH3‐assisted CO‐to‐CO2 dry oxidation protocol has been recently proposed as a clean and effective probe reaction for multinuclear Cu2+ species in Cu‐CHA catalysts.[ , , ] This reaction involves a two‐electron transfer process with an oxygen‐donor species, while the Cu2+‐to‐Cu+ reduction requires exchange of a single electron. Thus, dinuclear Cu2+ complexes necessarily result to catalyze CO oxidation; furthermore, ZCu2+(OH)− is the only Cu cation with a removable oxygen, suggesting it as the sole precursor of catalytic centers for CO oxidation. Dynamic simulations indicate the formation of such structures to occur during the NH3 adsorption phase: as mentioned earlier, the presence of NH3 ligands provides the Cu2+(OH)−(NH3)3 complex with the ability to inter‐cage diffuse. DFT has shown that, under these conditions, an exergonic pairing process is favored: two Cu2+(OH)−(NH3)3 can cohabit the same CHA cage in a two‐proximate configuration (Two‐P). The two main process steps can be summarized according to the following Equations (3) and 4:

The involvement of ZCu2+(OH)− ions has been verified by a second order kinetic analysis of CO oxidation on Cu‐CHA catalysts with variable SAR and Cu loading. Besides, the integral CO2 production analysis points to an expected asymptotic value which is in close agreement with the ZCu2+(OH)− population half. An independent UV‐Vis analysis verified how exposure of ZCu2+(OH)−‐rich catalysts to CO resulted in a significant reduction of the associated UV bands, while negligible changes were observed on a sample with a predominant population of Z2Cu2+. In fact, molecular dynamic simulations show that the Z2Cu2+ intra‐cage mobility is limited even in the presence of NH3; their reactivity has been observed only as a result of hydrolysis phenomena, not active here.[ , , ] The dry CO oxidation proceeds, therefore, according to a dual‐site mechanism involving isolated ZCu2+(OH)− species, precursors of the Two‐P complex, whose dynamic formation by NH3 solvation plays a critical role in the process.

The integral CO2 production is reported in Table 2. Using the values reported therein, it is possible to estimate the ZCu2+(OH)− reduction extent, (ZCu2+(OH)− red.) defined as twice the CO2 release with respect to the total ion population. Clearly, the presence of NH3 pre‐adsorbed on the catalyst enhances the reduction up to 3 times with respect to its absence (47.3 % vs. 16.4 %), due to the NH3 mobilizing effect. Noteworthily, even under fully NH3 saturated conditions, we never reached the complete site reduction within the 90 min duration of the phase. Indeed, the CO oxidation is a slow process, and the CO2 signal never approached the zero level under any of the considered conditions.

Table 2

Integral CO2 production and ZCu2+(OH)− fraction reduced by CO for different initial NH3 coverages (θNH3).

CO oxidation tests – variable θNH3 [%]	CO2 release [μmol]	ZCu2+(OH)− red. [%]
100	0.52	47.3
93	0.49	44.5
86	0.43	39.1
54	0.29	26.4
23	0.22	20.0
10	0.21	19.1
0	0.18	16.4

The dry CO oxidation protocol has been further replicated varying the NH3 adsorption time, as described in the Experimental Section. This procedure allows to assess the effect of different initial NH3 coverages on the CO oxidation activity, as well as the role of different adsorption sites. The new CO2 evolution data set is plotted in Figure 2. Limited NH3 loadings (θNH3=10 % and 23 %) result in CO2 profiles almost overlapping with the zero‐loading test. This is coherent with the corresponding NH3‐TPD curves, since at such low coverages, NH3 is expected to primarily bind to Brønsted acid sites, therefore being unable to mobilize the Cu2+ ions and to thus promote the CO‐to‐CO2 oxidation. A remarkable CO2 release increment is appreciated in correspondence to the Lewis feature onset at θNH3=54 %, ascribed to the progressive ZCu2+(OH)− solvation. At θNH3=93 %, we reach a profile close to the full saturation condition (θNH3=100 %), corresponding to a nearly entire mobilization of the cation population. The analysis of the ZCu2+(OH)− reduction extent leads to similar conclusions: from zero to low NH3 loadings, the estimated fraction is scarcely affected by the reactant presence (16.4–20.0 %), while a significant extent can be appreciated (26.4–47.3 %) concurrently to the coordination of the Cu2+ ions by NH3. These findings are consistent with the proposed dry CO oxidation protocol interpretation, and further validate the central role of NH3 in the formation of Cu2+ pairs (Two‐P) originating from ZCu2+(OH)−. Following the CO oxidation phase, a final NO+NH3 titration is performed to close the Cu2+ balance. The well‐established Cu2+ reduction chemistry over Cu‐CHA is represented by the reported reactions (Equations (5) and 6):[ , , , ]

The equimolar stoichiometries (Cu2+ : NO : N2=1 : 1 : 1) permit the direct quantification of the reduced Cu2+ by integrating the temporal NO consumption and N2 production traces. Table S2 reports these values, and a good match between the two integrals is always verified (NO/N2≈1). Remarkably, the reduction dynamics are clearly affected by the variable θNH3, as apparent in Figure 3. Recently, a RHC transients kinetic study on Cu‐CHA catalysts with pre‐adsorbed NH3 pointed out a second order reduction rate in the oxidized Cu2+ fraction. Comparable transients are observed here considering θNH3=100 %, where 500 ppm of N2 are detected upon co‐feeding NO and NH3, followed by a rapid decrease (Figure 3B); the NO consumption shows a mirror‐like trend (Figure 3A). Different initial NH3 loadings, however, determine a net profile alteration. Absent or scarce NH3 (θNH3=0–23 %) inhibits the reduction process, with a plateau feature corresponding to an approximately constant production of 100 ppm of N2. NO readily reacts with Cu2+‐NH3, as shown by both an independent spectroscopic study and our previous work on NH3 reactivity on Cu‐CHA; hence, the slow reduction observed here could be ascribed to the NH3 adsorption time required on the Cu2+ sites.[ , ] As the loading increases (θNH3=54–93 %), the reduction process results improved with a clear net N2 production peak, coherently with the wider fraction of solvated Cu2+ ions. Therefore, these trends show strong similarities with the CO2 evolution previously observed. The CO2 trace integrals (counted twice) and the averaged NO consumption/N2 production traces can be used to assess the extents of the two Cu2+ reduction processes, that is, by CO and by NO+NH3, combined with the knowledge of the sample total Cu2+ content. The estimated reduced Cu2+ fractions are reported in Table S2 and plotted against θNH3 in Figure 4 (Cu2+ red. by CO or by NO+NH3) alongside the overall Cu2+ balance, computed as their sum, which is closed with a maximum error of 2 %. The limited reduction extent by CO (8.2–10.0 %) al low NH3 coverages (θNH3=0–23 %) is verified by the NO+NH3 titration, which assesses a Cu2+ high oxidation state (89.3–92.0 %). In agreement with the previous considerations, the NH3 coordination on Cu2+ is crucial for the Cu‐CHA reduction processes; consequently, we observe relevant changes starting from a half coverage condition (θNH3=54 %), when Lewis sites begin to be active in the ammonia adsorption. Between θNH3=54 and 93 %, the Cu2+ fraction reduced by CO grows from 13.2 to 23.6 %, close to the full saturation conditions; correspondingly, the final NO+NH3 reduction titrates a constantly decreasing fraction of oxidized copper, nicely closing the Cu2+ balance.

Figure 3

NO (A) and N2 (B) evolution during the NO+NH3 feed, with different NH3 loadings. NO+NH3 feed: NH3=NO=500 ppm, CO=0 ppm, O2=0 % v/v, H2O=0 % v/v, balanced in He. GHSV=266250 cm3/h/gcat (STP) (Wcat=16 mg). T=200 °C. Pre‐oxidized catalyst.

Figure 4

Fraction of Cu2+ reduced by CO (gray) and by NO+NH3 (red) during CO oxidation tests with different NH3 loadings. The overall Cu2+ balance (black) is also shown.

CO oxidation – CO and NH3 co‐feed

As discussed, two phenomena concur during the NH3 adsorption phase, namely the saturation of the catalyst sites by NH3 and the formation of the Two‐P structure, that is, the active Cu2+ structure in the CO‐to‐CO2 oxidation. When co‐feeding NH3 and CO on a clean catalyst, the two processes proceed concurrently with the Two‐P reduction by CO, thus enabling a preliminary assessment of the relative rates of the reaction cascade. To prevent any Cu2+ reduction by gas phase NH3 and to obtain data directly comparable to the previous analysis, this test has been run applying a limited NH3 saturation time. Figure 5 depicts the CO2 evolution from the co‐feed experiment with 6 min of NH3 adsorption, compared to the reference test.

Figure 5

CO2 evolution during the CO+NH3 co‐feed test, where NH3 is removed after 6 min. CO+NH3 co‐feed: NH3=500 ppm (6 min), NO=0 ppm, CO=1000 ppm, O2=0 % v/v, H2O=0 % v/v, balanced in He. GHSV= 266250 cm3/h/gcat (STP) (Wcat=16 mg). T=200 °C. Pre‐oxidized catalyst.

Once CO and NH3 reach the catalyst, the co‐feed experiment trace shows an overlapping trend with the zero‐loading test (θNH3=0 %) during the very first minutes. Without ammonia, the mobility of ZCu2+(OH)− is limited and a negligible dimer formation, with a correspondingly slight CO oxidation activity, is expected. However, as the co‐feed test proceeds, the CO2 signal grows; indeed, during this phase, the Two‐P formation is favored due to the ammonia solvation effect and it becomes faster than the competing reduction by CO. A maximum is reached after 6 min of NH3‐feed (≈8 min), suggesting that the coupling of the Cu2+ ions is slower than the actual NH3 adsorption. Beyond this point, no more Two‐P formation is expected, the CO2 evolution results akin to the previous analyzed profiles and parallel to the trend collected with a 6 min NH3 pre‐adsorption phase, with comparable integral values (0.48 μmol). Similar findings have been observed when replicating the protocol with a 7 min feed of NH3 as shown in Figure S3. From this data set we propose the ammonia adsorption to be faster than the Two‐P formation process, while the CO oxidation turns out to be the slowest of the involved phenomena.

Conclusions

We have addressed the role of NH3 during the low‐T RHC‐SCR on a Cu‐CHA catalyst using transient response methods. A dedicated methodology has clearly shown that NH3 is adsorbed onto the Brønsted zeolite sites in the initial phase of the adsorption process, while the Lewis sites, extensively proved to be associated with the Cu2+ ions, become solvated at higher degrees of loading. Once coordinated with NH3 ligands, ZCu2+(OH)− becomes inter‐cage mobile and a thermodynamic driving force exists to enable its coupling with an adjacent complex in the same zeolite cage, forming a two‐proximate structure (Two‐P). This is regarded as the catalytic active site for the CO‐to‐CO2 oxidation, a clean probe reaction for the detection of dinuclear Cu2+ complexes. In this work, we have run dry CO oxidation tests preceded by a modulated NH3 adsorption, to replicate the loading and solvation conditions assessed in a previous NH3 adsorption study. Remarkably, significant CO2 production has been detected only when the Cu2+ sites were partially solvated (Lewis‐NH3), coherently with the formation of Two‐P complexes. The Brønsted‐NH3 instead was unable to catalyze the CO oxidation process. Similar features were observed during the following and final NO+NH3 titration phase: the presence of NH3‐Cu2+ is crucial to enable the reduction dynamics with the characteristic second order dependence on oxidized Cu2+ assessed in a dedicated transient kinetic study. Finally, based on CO+NH3 co‐feed experiments, we preliminarily propose that the NH3 adsorption process proceeds more rapidly than the Two‐P formation, both processes being faster than CO oxidation.

The present work further emphasizes the central role of NH3‐Cu2+ complexes in the low‐T SCR mechanism over Cu‐CHA and validates the use of dry CO oxidation as an effective transient method to titrate the formation of dinuclear Cu2+ complexes (Two‐P) during RHC conditions. The dual‐site nature of the Cu2+ reduction process corroborates the previously established mechanistic understanding for these systems.

Experimental Section

Catalyst characterization

The experimental campaign for this study was carried out on one research catalyst sample, a Cu‐exchanged chabazite (Cu‐CHA) in the form of powder, supplied by Johnson Matthey. The catalyst has been characterized in previous work by transient response methods and quantitative analyses. NO+NH3‐TPR, H2‐TPR and ICP‐MS analysis assigned a Cu loading of 1.7 % w/w (0.2750 mmol g−1 cat), confirming the equivalence between the amount of reducible Cu and the total Cu loading in the zeolite. The absence of CuOx species or unexchanged Cu was confirmed by UV‐Vis spectra. The SiO2/Al2O3 (SAR) and the Cu/Al ratios resulted respectively equal to 10 and 0.12. NO2 adsorption, followed by temperature programmed desorption, was used to evaluate a ZCu2+(OH)− fraction of 50 % (0.1375 mmol g−1 cat), further confirmed by deconvolution of the H2‐TPR profile.

Experimental set‐up

A quartz microflow reactor (internal diameter≈6 mm) was loaded with 16 mg of Cu‐CHA catalyst, diluted with cordierite up to 130 mg, with 90 microns as average particle size for both. The catalyst bed was suspended by means of quartz wool, preceded, and followed by quartz grains to enhance gas phase mixing. A K‐type thermocouple, directly in contact with the catalyst bed, was used to monitor the temperature evolution. The reactor was placed in a vertical electric oven, set in an experimental rig where Helium was used as balance gas. Mass flow controllers (Brooks Instruments) were used to dose the reactant gas species. Argon acted as tracer gas, and two fast 6‐way valves guaranteed step changes of NO, NH3, CO feed concentration, while the oxygen flow was regulated acting on the related mass flow controller. A saturator was used to feed water vapor and its concentration was adjusted by controlling the saturator temperature according to Antoine's Law. The temporal evolution of the gas species from the reactor outlet was followed by a UV+IR analyzer (ABB: LIMAS11 HW+URAS26) and a mass spectrometer (QGA Hiden Analytical), set in a parallel configuration, enabling the simultaneous measurement of all the species involved: CO, CO2, NH3, NO, NO2, N2, N2O, O2, H2O and Ar. The catalyst was initially hydrothermally preconditioned at 600 °C in the presence of 10 % v/v of O2 and H2O for 5 h to ensure catalyst cleaning. Prior to each test, the sample was treated at 550 °C with 8 % v/v of O2 for 1 h: this phase was used to remove previously adsorbed species and ensure complete catalyst oxidation. A cool‐down step, in the same gas feed, was realized to reach the operative temperature of the test. Three different experimental protocols were adopted for this study as illustrated in Figure S1, with a space velocity of 266250 cm3/h/gcat (STP).

Isothermal NH3 adsorption+TPD tests

Isothermal NH3 adsorption tests, with variable duration of the adsorption phase, followed by temperature programmed desorption (TPD), were used to study the interaction between NH3 molecules and the catalyst adsorption sites; the experimental protocol is depicted in Figure S1A. Ammonia adsorption was realized by feeding 500 ppm of NH3 in the presence of 8 % v/v of O2 at 200 °C, considering different injection time intervals: 1, 2, 4, 6, 7 min and up to complete catalyst saturation (1 h). The catalyst was then purged in O2 till NH3 concentration was below 5 ppm, to ensure the desorption of weakly bonded ammonia. A purge in He of 10 min only, with a cooling phase down to 190 °C, was used to guarantee complete O2 removal. During the following TPD part, the temperature was linearly increased in time up to 600 °C with a heating rate of 15°/min to release the strongly adsorbed NH3 molecules. An additional N2 release was observed associated with the Cu reduction by NH3.

Isothermal CO oxidation tests

The effect of variable NH3 loading on the formation of multinuclear Cu species was studied by isothermal CO oxidation tests with different adsorption times (Figure S1B). An initial step of isothermal NH3 injection, with 500 ppm of NH3 with 8 % v/v of O2, followed by an isothermal desorption, was realized. The same conditions were employed and adapted for the NH3 adsorption/desorption tests. Note that, when considering zero NH3 loading, this whole phase was neglected. A 15 minute He‐only purge was realized prior to the CO oxidation phase, where 1000 ppm of CO were fed for 90 min. Here, a transient CO2 release was recorded, associated with the multinuclear Cu2+ species reduction. Following a 30 minute purge, NO+NH3 titration was used to evaluate the residual Cu2+ fraction, co‐feeding 500 ppm of NO and NH3 for 1 h. During this phase, the N2 formation is observed, as the product of the Cu2+ reduction process. No production of further N‐species (N2O, NO2) was observed, as shown in a previous work. All the protocol steps were realized at 200 °C. Isothermal CO and NH3 co‐feed tests (Figure S1C) were realized to evaluate the competition between the multinuclear Cu2+ species formation and their reduction. After a 15‐minute purge in He, the reactor was co‐fed with 1000 ppm of CO, for 90 min, and 500 ppm of NH3, removed after 6 or 7 min. After a 30 minute purge in He, a final NO+NH3 titration was carried out co‐feeding 500 ppm of NO and NH3 for 1 h.

Supporting Information Summary

Detailed steps of the experimental protocol, material balances and the 7 minute NH3 feed Protocol C replica.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.