The selective catalytic reduction of NO_x by urea or ammonia (NH₃-SCR) is one of the most effective technologies to reduce NO_x emissions from diesel engine exhausts ^{[1, 2]}. Furthermore, the development of highly efficient and stable catalysts is crucial to meeting the upcoming strict NO_x emission standards ^[3]. Among the developed catalysts for the NH₃-SCR process, transition-metal-exchanged zeolites are promising candidates, showing attractive NH₃-SCR performance ^[4-6]. In particular, the Cu-CHA zeolites, including Cu-SSZ-13 and Cu-SAPO-34, have received considerable attention because of their outstanding NH₃-SCR activity over a wide temperature range, as well as their high hydrothermal stability ^[7-10]. In fact, Cu-SSZ-13 is used commercially by the BASF Corporation ^[8]. Compared to Cu-SSZ-13, Cu-SAPO-34 has unique characteristics: On one hand, Cu-SAPO-34 with a comparable Cu loading exhibits a higher activity at a low temperature range (150–200 ℃) ^{[11, 12]}, which is suitable for the start-up period of diesel vehicles. On the other hand, Cu-SAPO-34 is more robust to high-temperature steam ^[11]. In a typical after-treatment system for diesel engines, the SCR unit is normally used together with a diesel particulate filter (DPF), which requires periodic regeneration. A great deal of heat (> 700 ℃) is, thus, transferred to the SCR unit from the DPF ^[2]. Consequently, excellent hydrothermal stability is highly desirable for the NH₃-SCR catalyst. Finally and advantageously, the preparation cost of Cu-SAPO-34 is quite low. Moreover, the templates for SAPO-34 synthesis are abundant, which favors the manipulation of the features of SAPO-34 to fit reaction requirements.

One of the main problems facing Cu-SAPO-34 is its Cu-loading process, which generally requires time-consuming post-synthetic NH₄⁺ and Cu²⁺ ion-exchange processes in aqueous solution. During this process, the irreversible hydrolysis of Si–O(H)–Al bonds occurs, which results in structural damage and acidity and crystallinity loss, as well as the formation of CuO_x clusters, resulting in catalytic performance degradation ^[13]. Consequently, considerable efforts have been made to optimize the preparation process. Rare earth ions such as Ce³⁺ and La³⁺ have been used in the ion exchange process and are effective in mitigating dealumination and the aggregation of copper species^{[14, 15]}. A direct ion-exchange (DIE) method was developed by our group to achieve highly crystalline Cu-exchanged SAPO-34s ^[16]. In particular, calcination at high temperatures (700 ℃) was found to be critical for promoting the dispersion of Cu²⁺ ions. In addition, solid-state ion exchange (SSIE), calcination of SAPO-34/CuO mixtures at high temperatures and "one-pot" synthesis, where Cu-containing chemicals such as CuO and CuSO₄ are added during gel preparation, were developed by Gao et al. ^[17]. Unfortunately, despite the high-temperature calcination step to optimize the Cu²⁺ ion distribution, the NH₃-SCR catalytic activity remains insufficient because of the presence of CuO clusters.

Mirroring the development of molecular sieve synthesis ^[18], a "one-pot" synthesis of Cu-SAPO-34 using the copper-tetraethylenepentamine (Cu-TEPA) complex as a template and copper source has been developed ^{[19, 20]}. This method is convenient and facile. Cu²⁺ ions are formed after the removal of the organic part of the Cu-TEPA template, which is promising for practical applications. Unfortunately, the use of Cu-TEPA as a structure directing agent tends to result in Cu-SAPO-34 with high Si contents and abundant Si islands, resulting in inferior catalytic performance and hydrothermal stability. Moreover, it is challenging to optimize the Cu and Si contents and their distributions simultaneously. As a template, the Cu-TEPA complex has very strong directing ability toward the CHA structure. Even when using low doses of Cu-TEPA, which reduces the solid yield, the Cu content of the product cannot be decreased completely because of the preferential entry of copper into the CHA structure ^[19]. Thus, it would be desirable to find an effective way to control the Cu and Si contents and their distributions in this system. Martinez-Franco et al. improved the synthetic method by introducing various competitive templates ^[21]. A triple-template system was reported to maximize the solid yield and enhance the hydrothermal stability. However, activity decay was inevitable when the hydrothermal aging temperature was higher than 750 ℃ ^[21]. In addition, there were large Si islands in the products, which indicates that there is further scope for the optimization of the "one-pot" synthesis of Cu-SAPO-34 ^[22].

In this work, we propose a reconstruction strategy to synthesize Cu-SAPO-34 hydrothermally. In this strategy, Cu-rich SAPO-34 is prepared using Cu-TEPA as a precursor and copper source to assist the crystallization of Cu-SAPO-34. Because the role of Cu-TEPA as a template was restricted by the CHA structure and it is the only source of Cu, the synthesis of Cu-SAPO-34 could be controlled by choosing an appropriate template and adjusting the precursor dosage. Thus, Cu-SAPO-34 samples with similar Cu loadings but different Si contents or similar Si contents but different Cu loadings were synthesized and characterized using various techniques including magic angle spinning (MAS) solid-state ¹³C and ²⁹Si NMR, energy dispersive X-ray spectroscopy (EDX), electron paramagnetic resonance (EPR), in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), scanning electron microscopy (SEM), X-ray diffractometry (XRD), thermogravimetric and differential thermalgravimetric analysis (TGA and DTA, respectively), X-ray fluorescence (XRF) spectroscopy, N₂ adsorption–desorption, hydrogen temperature-programmed reduction (H₂-TPR), and ammonia temperature-programmed desorption (NH₃-TPD). The catalytic performance and hydrothermal stability of the compounds were evaluated based on their NH₃-SCR activities. The effect of the Si content (acidity) of the Cu-SAPO-34 samples on the catalytic performance was also investigated and is discussed.

Three batches of Cu-SAPO-34 precursors with minor compositional differences (Table 1), denoted Cu-SAPO-34-Pn (where n is the batch number), were prepared using a slightly modified version of the "one-pot" procedure reported in the literature ^[19]. The compositional modification was realized by changing the dosage of the silica source and template. The synthetic details for Cu-SAPO-34-Pn (n = 1–3) can be found in supporting information. The XRD patterns and typical morphologies of the Cu-SAPO-34-Pn samples are shown in Figs. S1 and S2, respectively.

Cu-SAPO-34 samples with different Cu contents were further synthesized by using the Cu-SAPO-34-Pn as the Cu source, part Si/P/Al source, and seeds. The synthetic procedure was similar to that of the Cu-SAPO-34-Pn samples, except for the omission of the Cu-TEPA template and the addition of Cu-SAPO-34-Pn. Pseudoboehmite (67 wt%) was first dissolved in distilled water, followed by the addition of phosphoric acid (85 wt%), silica sol (31 wt%), diethylamine (DEA), and Cu-SAPO-34-Pn. The molar proportions of the reactant gel were 0.9 P₂O₅/1.0 Al₂O₃/xSiO₂/1.9 DEA/55 H₂O (x= 0.4–0.7). The amount of Cu-SAPO-34-Pn precursor ranged from 17 wt%–60 wt%. The reactant gels were sealed in stainless-steel autoclaves and heated to 200 ℃ for 24 h with rotation. The samples were recovered by filtration, washed with distilled water, and dried at 120 ℃ in air overnight. Before the catalytic tests, the samples were calcined at 600 ℃ for 3 h to remove the templates. More details are listed in Table 1.

Elemental analysis of the samples was carried out using a Philips Magix-601 XRF spectrometer. The powder XRD patterns were recorded on a PANalytical X'Pert PRO X-ray diffractometer with Cu-K_α radiation (λ = 1.54059 Å) generated at 40 kV and 40 mA. The textural properties of the calcined samples were determined by N₂ adsorption at –196 ℃ on a Micrometrics ASAP 2020 system. The total surface area was calculated using the Brunauer-Emmett-Teller (BET) equation, and the micropore volume and surface area were determined by using the t-plot method. DRIFT spectra were recorded on a Varian Cary-5000 UV-Vis-near infrared (NIR) spectrophotometer equipped with an integrating sphere, and the background of the UV-Vis spectrum was plotted against a BaSO₄ reference standard. The morphology of the samples was observed by SEM on a Hitachi SU8020 field-emission (FE) SEM device. EDX measurements conducted on the same Hitachi SU8020 FE-SEM device equipped with a Horiba X-Max silicon drift X-ray detector. The ¹³C and ²⁹Si MAS NMR spectra were recorded on a Bruker Avance Ⅲ 600 spectrometer equipped with a 14.1-T wide-bore magnet. The resonance frequencies were 100.5 and 119.2 MHz for ¹³C and ²⁹Si, respectively. The spinning rates of the samples at the magic angle for ¹³C and ²⁹Si were 8 and 6 kHz, respectively. The reference material for the chemical shift (in parts per million (ppm)) determination for ¹³C and ²⁹Si was the sodium salt of 2, 2-dimethyl-2-silapentane-5-sulfonate (DSS). TGA and DTA were performed on a TA SDTQ600 analyzer at a temperature programmed rate of 10 ℃/min under an air flow of 100 mL/min. The organic contents of the as-made materials were determined by elemental analysis performed with an Elementar Vario EL elemental analyzer.

The NH₃-TPD experiments were conducted in a Micromeritics Autochem Ⅱ 2920 device. In these tests, 100 mg of calcined sample particles (40–60 mesh) were loaded into a U-shaped quartz tube and pretreated at 650 ℃ for 1 h in He flow (30 mL/min) to remove any impurities and, then, treated in a gas mixture of 2% NH₃ and 98% He flow (30 mL/min) at 120 ℃ to saturate the sample surface with adsorbed NH₃ (60 min). Subsequently, He (30 mL/min) was purged through the sample for 30 min to remove the weakly adsorbed NH₃ molecules. The measurement of the desorption of NH₃ was performed from 100 to 600 ℃ (10 ℃/min) under He flow (30 mL/min).

DRIFT spectra were recorded with a Bruker Tensor 27 Fourier transform infrared (FTIR) spectrometer equipped with a gas cell. For this analysis, 15 mg of calcined sample was loaded into the sample cup of the high-temperature reaction chamber. For the ammonia-adsorbed in situ DRIFTS experiments, the pretreatment was conducted at 550 ℃ in N₂ flow (20 mL/min) for 1 h before cooling to 150 ℃. NH₃ adsorption was conducted with a 24 mL/min flow rate containing 0.083% NH₃ with N₂ balance for 30 min, and then the DRIFTS spectra were obtained at different temperatures (150–500 ℃) in a flow of pure N₂ flow (20 mL/min). The in situ DRIFTS spectra were also measured during reaction at 200 ℃. The preadsorption of NH₃ was similar to that described above but was carried out at 200 ℃. After that, the sample was swept with pure N₂ flow (20 mL/min) for 20 min, and then treated with 0.083% of NO and 6.1% of O₂ with N₂ balance for 0–30 min. The DRIFTS spectra were obtained during this process. The spectra were treated by subtracting the background spectra recorded after exposing the sample to 20 mL/min pure N₂ at 200 ℃.

EPR measurements were performed on a Bruker A200 spectrometer. The samples were sealed in quartz tubes and pretreated at 120 ℃ in pure N₂ flow for 12 h. During spectral collection, the microwave power was 2 mW, and the frequency was 9.52 GHz. The sweep width was 2000 G, and the sweep time was 84 s, modulated at 100 kHz with a 2 G amplitude. A time constant of 41 ms was used. The spectra were collected at 102 K. For the quantification of isolated Cu²⁺ species, copper sulfate solution was used as a standard (at 102 K).

The temperature-programmed reduction of hydrogen (H₂-TPR) was measured on a Micromeritics Autochem Ⅱ 2920 device. In the TPR experiments, 100 mg of calcined sample particles (40–60 mesh) was pretreated in Ar flow (30 mL/min) at 500 ℃ for 1 h to remove impurities and, then, cooled to 100 ℃ in Ar flow (30 mL/min). The H₂-TPR experiments were carried out at a heating rate of 10 ℃/min from 100 to 900 ℃ under a 10% H₂ and 90% Ar flow (30 mL/min), and the consumption of H₂ was determined using a thermal conductivity detector (TCD).

The catalytic activity for the NH₃-SCR was determined using 60 mg of the Cu-SAPO-34 catalyst (60–80 mesh) mixed with 240 mg of quartz beads (60–80 mesh) in a fixed-bed quartz reactor (0.95-cm diameter), and a K-type thermocouple was placed in the center of the catalyst bed to determine the temperature of the catalyst. The reactant gases were regulated before entering the reactor using mass flow controllers (Brookers). All gas lines were heated to 100 ℃ to prevent the condensation of the vapor. The concentrations of the simulated gases were as follows: 0.05% NO, 0.05% NH₃, 6.1% O₂, 6.4% H₂O, and N₂ balance. The total flow rate was 320 mL/min, and the corresponding gas hourly space velocity (GHSV) was 300000 h^–1. The concentrations of NO_x (NO and NO₂) and N₂O in the inlet and outlet gases were continually analyzed by FTIR (Tensor 27, Bruker) in a 2-m gas cell. All catalysts were pretreated in the simulated gases at 550 ℃ for 2 h before the activity tests. Catalytic activity tests were carried out over the temperature range of 150 to 500 ℃, and the activity results were recorded after holding at each temperature for 40 min. The NO_x conversion was calculated using Eq. (1).

(1)

The N₂ selectivity was calculated using Eq. (2).

(2)

NH₃-SCR kinetic tests were performed in the same quartz tube reactor using 16.5 mg of the catalyst mechanically mixed with 300 mg of quartz sand. The total flow rate was 930 mL/min, corresponding to a GHSV of 3300000 h^-1. The turnover frequencies (TOF) were calculated using eqn. (3).

(3)

Here, F is the flow rate of NO_x, and M is the number of moles of Cu in the powdered catalyst.

After the standard NH₃-SCR evaluation, the samples were further heated to 600 ℃ in dry air for ca. 15 min in the quartz reactor. Subsequently, gas flow containing 10% water balanced with N₂ was passed through the sample at 800 ℃ for 16 h to achieve high-temperature hydrothermal treatment. The total flow rate was 253 mL/min. Before any characterization or evaluation, the steam-treated samples were further heated to 600 ℃ in dry air and then kept in an oven at 120 ℃.

Cu-SAPO-34 samples with varied Si and Cu contents were synthesized by the reconstruction strategy. The crystallization conditions, product compositions, and solid yields are listed in Table 1. It was found that the Cu content could be adjusted by altering the dosage of the Cu-SAPO-34-Pn precursor (samples 1–3). Cu-SAPO-34 with a very low Cu loading can be directly synthesized in high solid yield by introducing a small amount of Cu-SAPO-34-Pn, as seen for sample 1, LSiCu_1.6 and HSiCu_1.7, for which direct synthesis using the Cu-TEPA template is challenging. The Cu content increased as the amount of precursor increased (samples 1–3). This result indicates that the precursor acts as an effective copper source in the crystallization process. The Si contents of sample 4 and Cu-SAPO-34-P1 were similar when the dosage of xSiO₂ was 0.6. This result suggests that the newly crystallized SAPO-34 should have a Si content comparable to that of the precursor under these reaction conditions. When the amount of xSiO₂ was reduced or increased while the other conditions remained constant, the Si contents decreased or increased, respectively, as shown for samples 2, 4, and 5. It should be noted that the Si content of the product is affected by the composition of precursor significantly, as seen for sample 5 and HSiCu_3.5. Although the dosage of xSiO₂ for HSiCu_3.5 is lower than that of sample 5, the Si content of HSiCu_3.5 is higher because of the usage of Cu-SAPO-34-P3 with higher Si contents than Cu-SAPO-34-P1. On the basis of these results from the reconstruction process, four Cu-SAPO-34 samples with comparable Cu and Si contents were synthesized (denoted LSiCu_1.6, HSiCu_1.7, LSiCu_3.4, and HSiCu_3.5) and used for further study. L and H refer to low and high product Si contents respectively, and the number following Cu is the percentage Cu content as determined by XRF analyses.

The XRD patterns of LSiCu_3.4 and HSiCu_3.5 are displayed in Fig. 1. All the peaks can be indexed to the typical CHA structure. There are no peaks at 35.3° and 38.5°, suggesting the absence of CuO species and the advantage of the current synthesis in achieving a uniform Cu distribution. N₂ adsorption measurements were conducted, and the isotherms are shown in Fig. 2. Both LSiCu_3.4 and HSiCu_3.5 show typical type-Ⅰ isotherms and have microporous surface areas of 521 and 535 m²/g, respectively. The micropore volumes were 0.26 and 0.28 cm³/g (Table S1), confirming their good crystallinity.

UV-Vis spectroscopy was used to identify the copper species in the products, and the spectra are shown in Fig. 3. Both spectra contain strong bands at 270 nm, comparable to that of Cu-SAPO-34-P2. This result indicates that the Cu species in the as-synthesized products were maintained, i.e., as the Cu-TEPA complex ^[19]. SEM and EDX analyses were performed to visualize the product morphology and Cu distribution, respectively, in the calcined products. Both LSiCu_3.4 and HSiCu_3.5 have typical rhombohedral morphologies with similar particle sizes of ca. 3–5 μm (Fig. 4). The EDX mapping images of Cu for LSiCu_3.5 and HSiCu_3.4 show highly dispersed light spots, as seen in the inset of Fig. 4. These results confirm that the precursor participates in the crystallization and the Cu species (Cu-TEPA complex) were incorporated into the Cu-SAPO-34 products uniformly.

Solid-state ¹³C MAS NMR was employed to investigate the incorporated organic templates. The peak assignments are shown in Fig. 5. The peaks at 10 ppm are ascribed to the –CH₃ groups of DEA, and the peaks at around 41 ppm are obviously widened, probably arising from the combination of two signals from the –CH₂– groups of DEA and those from the –CH₂– groups linked to the primary amine groups of TEPA. The peaks at 48 ppm are assigned to –CH₂– groups connected to the secondary amine groups of TEPA ^{[23, 24]}. The results imply that both DEA and Cu-TEPA are still present in the samples. The exact molar ratio of TEPA/DEA was further estimated using TGA/DTA and elemental analyses. The results are listed in Table 2 and Fig. S4. Because the C/N ratio of DEA is 4 and that of TEPA is 1.6, the molar percent of TEPA in the full templates was calculated to be 21%; for comparison, 75 wt% content is obtained using the "one-pot" synthesis of Cu-SAPO-34 ^[21]. The obviously decreased Cu-TEPA content implies that the role of Cu-TEPA as a template was restricted by the CHA cage.

The chemical environments of the Si atoms in LSiCu_3.4 and HSiCu_3.5 were investigated using solid-state ²⁹Si MAS NMR, and the spectra are shown in Fig. 6. In the spectrum of LSiCu_3.5, only a single signal at –91 ppm was observed, and this is attributed to the Si(4Al) species ^[25]. It is not surprised as the SAPO-34 directed by DEA template always has abundant Si(4Al) environments ^[26]. The ²⁹Si NMR spectrum of HSiCu_3.4 is similar, but there is a small peak at –110 ppm. The presence of this peak suggests that a very small number of Si islands were generated as the silica content increased. Nevertheless, the Si content and coordination environment have been better optimized compared to previous works ^{[19, 21]}. The isolated Si(4Al) coordination environment in the reconstructed Cu-SAPO-34 samples is beneficial for framework stability and the dispersion of Cu²⁺ ions ^{[7, 27]}. A better NH₃-SCR activity is therefore expected which will be discussed in the following part.

The NH₃-SCR activities of LSiCu_3.4 and HSiCu_3.5 were evaluated under standard NH₃-SCR conditions at a GHSV of 300000 h^–1. The reaction results are presented in Figs. 7a and S5. Both catalysts gave ca. 100% N₂ selectivity over the whole reaction period. The LSiCu_3.4 sample gave a high NO_x conversion of 51% at 150 ℃, which rapidly increased to 95% at 175 ℃, further increasing to 100% at 200 ℃. The NO_x conversion was maintained at 100% until 350 ℃. On further increasing the reaction temperature to 400 ℃, the conversion dropped to 94%, then decreasing to 89% at 500 ℃. Compared to LSiCu_3.4, HSiCu_3.5 shows an obviously lower NO_x conversion at low temperatures (150 ℃, 40.8% and 175 ℃, 87.7%) but better performance at high temperatures (500 ℃, 96%). The results suggest that, at an equal Cu loading level, the relatively low Si content of Cu-SAPO-34 may have a positive effect on the NH₃-SCR activity at low temperatures, whereas the high Si content benefits the activity at high temperature. This illustrates that the Si content (acidity) of the catalysts has a significant influence on the NH₃-SCR reaction. This hypothesis was further confirmed on the basis of the results obtained with LSiCu_1.6 and HSiCu_1.7, whose catalytic performances show a similar trend (Fig. 7b). It is interesting to note that the activities of LSiCu_1.6 and HSiCu_3.5 are comparable, as shown in Fig. S6. These results demonstrate the importance of both Si and Cu species of Cu-SAPO-34 catalyst for the NH₃-SCR reaction.

Two kinds of reported Cu-SAPO-34 samples were used for comparison with our samples. The reported "one-pot" synthesized Cu-SAPAO-34 (SAPO-34-7, 0.178 Si/TO₂, 3.1 wt% Cu) with Cu-TEPA as the Cu source and co-template showed an obviously lower NO_x conversion after hydrothermal treatment at 750 ℃ for 16 h, and the highest NO_x conversion decreased from ca. 95% to 70% ^[21]. The ion-exchanged Cu-SAPO-34 (0.167 Si/TO₂, 3.97 wt% Cu) ^[1] had lower catalyst activity, and its highest NO_x conversion was ca. 90% (GHSV = 240000 h^–1). However, the NO_x conversion was maintained well after steam treatment at 800 ℃ for 16 h. For comparison, we also prepared a "one-pot" Cu-SAPO-34 reference sample using Cu-TEPA and DEA as co-templates (denoted RMSiCu_3.5). This sample had comparable Cu contents but medium Si contents. Although a 10 wt% nanosized SAPO-34 seed was used to induce the crystallization of RMSiCu_3.5, the crystallinity (Table S1) and solid yield (40.0 wt%) were still low. As shown in Fig. 7, the NO_x conversion of RMSiCu_3.5 lies between those of LSiCu_3.4 and HSiCu_3.5, but it shows more serious deterioration in activity upon hydrothermal treatment at 800 ℃ for 16 h. The loss in activity for LSiCu_3.4 and HSiCu_3.5 are less than 10%, whereas the loss in activity reached 24% for RMSiCu_3.5 (Table 3). The results demonstrate the better hydrothermal stability of the reconstructed Cu-SAPO-34 and is comparable to that of ion-exchanged Cu-SAPO-34 ^[11].

Kinetic measurements were carried out to determine the TOF and apparent activation energy (E_a) and evaluate the reason for the activity difference between the reconstructed and "one-pot" synthesized samples in the NH₃-SCR reaction. For the kinetic tests, a GHSV of 3300000 h^–1 and pellet diameter of 60–80 mesh were employed to eliminate the influence of diffusion limitations ^[28]. Fig. 8 shows the Arrhenius plots of LSiCu_3.4, HSiCu_3.5, and RMSiCu_3.5, and the plots were used to calculate the E_a values. LSiCu_3.4 has the smallest E_a value of the three fresh samples, suggesting that it has the lowest activation barrier for the NH₃-SCR in the temperature range of 150–210 ℃. After hydrothermal treatment, the TOF values of the samples decreased as the E_a values rose. The RMSiCu_3.5-HT sample presented the most significant increase in the E_a value, which is obviously higher than those of the other two samples, indicating the superior NH₃-SCR performance and hydrothermal stability of the reconstructed Cu-SAPO-34 samples.

The NH₃-TPD profiles of the samples are shown in Fig. 9. Each desorption curve can be decomposed into three peaks centered at ca. 190 ℃ (arising from weak acid sites from surface hydroxyl groups), 290 ℃ (arising from moderate acid sites, i.e., both Lewis acid sites related to Cu species and structural Brönsted acid sites), and 440 ℃ (assigned to strong Brönsted acid sites) ^[29]. The TPD profile of HSiCu_3.5 contains a larger desorption peak area and higher NH₃ desorption temperature than those of LSiCu_3.4, suggesting that it has more acid sites and stronger acidity. The NH₃-TPD process was further investigated using in situ DRIFTS difference spectra. In Fig. 10, the negative band at 3673 cm^–1 can be assigned to P-OH sites occupied by NH₃, and the bands at 3621 and 3600 cm^–1 are caused by the depletion of Si-OH-Al acid sites by NH₃ ^{[30, 31]}. The bands at 3271 and 1459 cm^-1 can be assigned to the stretching and bending vibrations of NH₄⁺ species adsorbed on Brönsted acid sites, and the bands at 3184 and 1278 cm^–1 are attributed to NH₃ adsorbed on Lewis acid sites. As the temperature was increased, NH₃ desorption occurred, resulting in a decrease (or even disappearance) in the band intensity. The bands at 3621, 3600, 3271, and 1459 cm^-1 became very weak at 400 ℃ for LSiCu_3.4, but they are still distinct in the spectrum of HSiCu_3.5. This result again indicates again that the Brönsted acid sites of HSiCu_3.5 have stronger adsorption capacity than those of LSiCu_3.4.

Because a higher number of acid sites can better inhibit the NH₃ oxidation reaction and improve the NO conversion at high temperatures ^{[29, 32]}, the observation of improved high-temperature NH₃-SCR catalytic performance on HSiCu_3.5 is understandable. However, the reason for its inferior low-temperature NH₃-SCR activity merits investigation. Wang et al. ^[22] reported that a lower number of Brönsted acid sites results in fewer available NH₄⁺ species, thus resulting in lower NH₃-SCR activity. However, the acid strength was not considered in their research. Li et al. ^[33] found NH₃ migration is slower than the NH₃-SCR reaction at low temperature range. Thus, the stronger Brönsted acidity of HSiCu_3.5 may cause slower NH₃ migration to the active Cu sites, which is disadvantageous for the NH₃-SCR reaction. To verify this speculation, samples with preadsorbed NH₃ were exposed in NO-O₂ gas at 200 ℃ and monitored by in situ DRIFTS. As shown in Figs. 11a and 11b, all the peak intensities decrease with prolonged exposure time. The peak intensity at 3271 cm^–1, which represents the amount of NH₄⁺ at Brönsted acid sites ^[31], decreases slowly for HSiCu_3.5 (Fig. 11c), verifying the lower activity of NH₃ adsorbed on the Brönsted acid sites of HSiCu_3.5.

It is well accepted that Cu species play a crucial role in the NH₃-SCR reaction. EPR measurements were conducted because EPR can be used to quantify isolated Cu²⁺ ions while not detecting species such as CuO, Cu⁺, and [Cu-O-Cu]^{2+ [34]}. The EPR spectra of LSiCu_3.4 and HSiCu_3.5 are similar, as shown in Fig. 12, and only one type of Cu species with signals of g_∥ = 2.36 and A_∥ = 139 G was identified. This result implies that all samples have only one type of isolated Cu²⁺ species located at the double six-membered-rings (site I) ^{[28, 35]} under the current measurement conditions, and steam treatment had a limited effect on the location of Cu²⁺ ions. The isolated Cu²⁺ ion contents were estimated from the EPR spectra, and the results are listed in Table 4. The calculated Cu contents are lower than those obtained from the XRF results because of the existence of other Cu species ^{[34, 36]}. HSiCu_3.5 has more isolated Cu²⁺ ions than LSiCu_3.4. Given their similar Cu loadings, we can infer that the larger number of acid sites on HSiCu_3.5 is beneficial for stabilizing the isolated Cu²⁺ ions. It has been recognized that a higher isolated Cu²⁺ ion content is conducive to the NH₃-SCR catalytic activity at low temperatures ^[36]. Therefore, the lower catalytic activity of HSiCu_3.5 at low temperatures compared to that of LSiCu_3.4 implies that the acidity of Cu-SAPO-34 plays a more important role than expected in the low-temperature NH₃-SCR activity. The isolated Cu²⁺ contents increase at a comparable level for the investigated samples after high-temperature steam treatment, indicating that the high-temperature treatment has a beneficial effect on the distribution of Cu²⁺ ions.

H₂-TPR was employed to probe the reducibility of the Cu species. The TPR profiles of the three samples present two main reduction peaks at ca. 240 and 600 ℃, as seen in Fig. 13. The low-temperature peak is attributed to the reduction of isolated Cu²⁺ to Cu⁺, and the high-temperature peak is due to the further reduction of Cu⁺ to Cu^{0 [37-40]}. The small shoulder peaks at 175 ℃ for HSiCu_3.5 and RMSiCu_3.5 represents the reduction of nanosized CuO to Cu^{0 [39]}. After hydrothermal treatment, the high-temperature peaks moved to 700–800 ℃, and their intensities decreased obviously, implying a more difficult reduction for Cu⁺ ions. This is likely owing to the better redistribution of isolated Cu²⁺ ions during aging. Moreover, a new peak at ca. 450 ℃ appeared for aged RMSiCu_3.5 (sample RMSiCu_3.5-HT), suggesting the formation of bulk CuO species. Comparatively, the reconstructed samples have better resistance to steam treatment, showing little change in Cu species as compared to RMSiCu_3.5. This is consistent with the better catalytic performance of the aged reconstructed samples.

A new reconstruction strategy has been developed to synthesize Cu-SAPO-34 catalysts, which offers a wide crystallization phase region and allows an easy control over the Si and Cu contents. The resultant Cu-SAPO-34 has a dominant distribution of Si(4Al) coordination sites due to the restricted template role of Cu-TEPA in the precursor. Excellent NH₃-SCR catalytic performance and enhanced resistance to hydrothermal conditions were achieved using the Cu-SAPO-34 catalyst prepared by the reconstruction method. Our investigation revealed that the acidity of Cu-SAPO-34 has a great effect on the NH₃-SCR activity and hydrothermal stability. The weaker acidity is beneficial for low-temperature NH₃-SCR activity. Therefore, the control of the Si content and distribution are crucial for catalyst optimization. These results suggest that Cu-SAPO-34 catalyst prepared by the reconstruction method is a promising candidate for the NH₃-SCR process.