With a heightened sense of awareness of the environment, emission regulations have become extremely stringent. NO_x can contribute to the formation of photochemical smog, acid rain, haze etc, Thus, one of the most important aspects of these regulations is to lower NO_x emissions [1, 2]. Until now, selective catalytic reduction by ammonia (NH₃-SCR) was considered an effective technique to eliminate NO_x in the lean-burn condition [1, 2]. In a typical diesel emission after-treatment system, diesel particle filters (DPFs) are generally placed beneath the NH₃-SCR catalyst to trap particulate matter (PM). The regeneration process requires the DPFs to be heated to a high temperature (usually > 873 K) [3]. Moreover, the SCR catalyst is often exposed at low temperatures (usually < 373 K) during the diesel storage and its cold-start period. Thus, the NH₃-SCR catalyst needs to have both excellent high-temperature (HT) and low-temperature (LT) hydrothermal stabilities and catalytic performance.

Thus far, Cu-exchanged zeolites, such as Cu-beta (Framework type code (FTC): *BEA) [1, 2, 4], Cu-ZSM-5 (MFI) [1, 2, 5], Cu-SSZ-13 (CHA) [1-3, 6], Cu-SSZ-39 (AEI) [1, 7], Cu-LTA (LTA) [8], Cu-SSZ-16 (AFX) [9], and Cu-SAPO-34 (CHA) [1, 3], have been considered as promising NH₃-SCR catalysts. Among these, silicoaluminophosphate (SAPO)-based catalysts are attracting increasing attention owing to their excellent HT hydrothermal stability. To our knowledge, among the 41 SAPO molecular sieves (MSs) [10], only Cu-SAPO-34 (CHA), Cu-SAPO-18 (AEI) [11], Cu-SAPO-35 (LEV) [12], MnO_x-SAPO-11 (AEL) [13], and Cu-SAPO STA-7 (SAV) [14] have been used as NH₃-SCR catalysts. With the exception of SAPO-11, the others have similar structural features: small pore openings and large cages that show promising NO_x conversion. However, SAPOs are known to exhibit LT hydrothermal instability. Therefore, it motivates researchers to explore other small pore sieves with both HT and LT hydrothermal stability characteristics. Apart from the aforementioned small pore SAPO MSs, SAPO-17 (FTC: ERI) is another small pore cage-based MS that might be a potential NH₃-SCR catalyst.

SAPO/AlPO-17 has a three-dimensional (3D) 8 × 8 × 8 ring channel system [15]. It consists of eri-cage (Fig. 1a) and can-d6r columns (Fig. 1b) running along the c-axis. Each can-d6r column links with six similar columns, creating large ellipsoidal eri cages (6.79 × 12.07 Å) (Fig. 1c). SAPO/AlPO-17 can be synthesized using a variety of organic structure directing agents (OSDAs), such as piperidine (PI) [15], quinuclidine [16], cyclohexylamine (CHA) [17-19], neopentylamine [20], methylamine [21], N, N, N', N'-tetramethyl-1, 6-hexanediamine (TMHD) [22], N, N, N, N', N', N'-hexamethyl hexamethylene hydroxide [23], and 1, 4-bis (N-methylpyrrolidinium) butane hydroxide [24].

Knowledge of the locations of OSDAs within cages and the host (framework)-guest (OSDA) interactions can provide indirect information regarding Brönsted acid sites [25]. Additionally, this knowledge might provide the possible locations of exotic active sites such as Cu²⁺ for NH₃-SCR. Molecular modelling [26], spectroscopic characterization (UV-Raman, NMR, and IR) [27-29], and refinement against diffraction data (X-ray, electron, and neutron) [25, 30, 31] can be used to determine the locations of the guest species. Compared with other methods, Rietveld refinement against powder X-ray diffraction (PXRD) not only provides the atomic coordinates of guest species and the host framework, but also considers the effects arising from the synthesis condition (such as different types of silica precursors). For example, the structural features of OSDA directing DNL-6, a novel small pore RHO-type SAPO material, have been explored using Rietveld refinement; subsequently, the targeted synthesis of DNL-6 using a series of predicted OSDAs has been accomplished [32]. Therefore, Rietveld refinement is an appropriate method to investigate the locations of OSDAs in SAPO/AlPO-17. Thus far, Joseph J. Pluth and Alain Tuel et al have utilized single-crystal X-ray diffraction (SXRD) to investigate the precise locations of PI and TMHD in AlPO-17 [15, 22]. The exact positions of other OSDAs in SAPO/AlPO-17 remain undiscovered.

In this study, we synthesized SAPO-17 with tunable Si contents (the samples are denoted as SAPO-17-m, where m indicates Si/(Si+Al+P)) by using commercially available, inexpensive CHA as OSDA, and we investigated their locations, host-guest interactions, and Brönsted acid sites through Rietveld refinement against PXRD data. Moreover, we explored the NH₃-SCR activities of Cu-SAPO-17 with different Si and Cu contents; the catalysts were denoted as Cu-SAPO-17-m-n, where m and n indicate Si/(Si+Al+P) and Cu/Si, respectively. Among these catalysts, Cu-SAPO-17-8.0%-0.22 displays favorable catalytic performance and HT and LT hydrothermal stability characteristics.

Pseudo-boehmite (66.5%), aluminium isopropoxide (Aladdin, ≥ 98%), phosphoric acid (80%), Fumed silica (FS, CAB-O-SIL M-5), colloidal silica (CS, 27.34%), tetraethylorthosilicate (TEOS, Kermel, ≥ 98%), cyclohexylamine (CHA, Sinopharm Chemical Reagent, ≥ 99%), aluminum nitrate (Sinopharm Chemical Reagent, ≥ 99%), sodium hydroxide (Aladdin, ≥ 99%), potassium hydroxide (Damao Chemical Reagent Factory, ≥ 85%), potassium chloride (Kermel, ≥ 99%), and hydrofluoric acid (Damao Chemical Reagent Factory, 40%) were used as received. 1, 4-bis (N-methylpyrrolidinium) butane bromide (1, 4-MPBr₂) was synthesized according to reference [33].

Fumed silica was added to the solution containing aluminum nitrate, sodium hydroxide, potassium hydroxide and 1, 4-MPBr₂. When the mixture formed a homogeneous gel with the composition of 30 SiO₂: 0.5 Al₂O₃: 15 NaOH: 15 KOH: 4.5 1, 4-MPBr₂: 1191 H₂O, it was transferred to a stainless-steel autoclave and crystallized at 433 K for 110 h. The product was obtained by centrifugation, washed with deionized water, and dried at 373 K overnight. The heterogeneous seeds were obtained by calcinating the as-made sample at 873 K for 4 h and exchanged with 1 M NH₄NO₃ solution at 353 K for 2 h (repeat three times).

SAPO-17 was synthesized according to the previously reported procedure [34, 35]. The gel compositions are listed in Table S1, and the gel is prepared by the following procedure. An aluminium source (pseudo-boehmite or aluminium isopropoxide) and phosphoric acid were mixed with deionized water for 2 h, and a silica source (colloidal silica or aluminosilicate zeolite), CHA, HF, and KCl were added to the mixture stirred until the gel was homogeneous. Finally, seeds were added to the gel and stirred for another 30 min. The mixture was transferred to a stainless-steel autoclave and crystallized at 473 K for a known time. In all cases, the products were obtained by centrifugation, washed with deionized water, and dried overnight at 373 K.

Cu-SAPO-17 was prepared according to the previous reported method [36]. The as-made samples were exchanged with appropriate Cu in the aqueous solution of cupric acetate monohydrate at 353 K for 4 h. The Cu contents of Cu-SAPO-17 were measured by XRF and the results are listed in Table 1. When testing the NH₃-SCR activities, Cu-SAPO-17 samples were calcined at 873 K for 2 h.

The powder X-ray diffraction (PXRD) data for phase identification was collected on a PANalytical X'Pert PRO X-ray diffractometer (Cu K_α, λ = 1.5418Å). PXRD data (2θ ranges from 5° to 120°) used for the Rietveld refinement was recorded on the high-resolution STOE STADI P ESSENTIAL diffractometer equipped with a Mythen ll detector in the Debye-Scherrer mode (Cu Kα₁, λ = 1.5406 Å). The diameter of the capillary is 0.2 mm. Scanning emission microscope (SEM) images were performed on a Hitachi SU8020 microscope. PANalytical Axios advanced X-ray fluorescence spectroscopy (XRF) was used to determine the content of Si, Al, P, Cu, Na, and K. Thermogravimetry analysis (TGA) was performed on a TA Q-600 analyzer at a rate of 10 K/min from room temperature to 1173 K.

The in-situ diffuse reflectance infrared Fourier transform (DRIFT) spectra were measured on a Bruker Vertex 70. Undiluted samples were used to investigate the vibrations of CHAs. To eliminate the effect of adsorbed water prior to collecting the spectrum at 473 K in a N₂ stream, the samples were dried at 473 K for 1 h in the presence of N₂ (50 mL/min).

The NH₃-IR spectra were recorded on Bruker Tensor 27 spectrometer in the frequency range 4000‒800 cm^‒1 with a resolution of 4 cm^‒1. Prior to the experiments, catalysts were heated to 773 K under vacuum and maintained at this temperature for 1 h. Eventually, they cooled down to 423 K and adsorbed NH₃ at this temperature for 30 min. Finally, spectrum was recorded after desorbing NH₃ at 473 K for 40 min.

The N₂ adsorption experiments were performed on Micrometrics ASAP 2020 analyzer. Prior to the test, samples were degassed at 623 K for 4 h. NH₃-TPD was recorded on Micromeritics AutoChedm II chemisorption analyzer. Before the test, samples were heated to 873 K and maintain at this temperature for 30 min in the presence of He. Then samples were cooled down to 423 K, adsorbed NH₃ at this temperature for 30 min, and purged by He for another 30 min. Finally, samples were heated to 923 K with a rate of 10 K/min and TCD was used to detect the signals.

The solid-state MAS NMR experiments were recorded on a Bruker Avance lll 600 spectrometer equipped with a 14.1 T wide-bore magnet using a 4 mm WVT probe. The resonance frequencies of ¹H, ¹³C, ²⁹Si, ²⁷Al, and ³¹P are 600.13 MHz, 150.9 MHz, 119.2 MHz, 156.4 MHz, and 242.93 MHz, respectively. ¹³C MAS NMR spectra and ²⁹Si MAS NMR spectra were recorded using cross-polarization (CP, ¹H-¹³C, and ¹H-²⁹Si) sequence at the spinning rate of 12 kHz and 8 kHz, respectively. The chemical shifts were referenced to adamantine and 4, 4-dimethyl-4-silapentane sulfonate sodium salt at 29.5 ppm, 0 ppm, respectively. ²⁷Al MAS NMR and ³¹P MAS NMR were recorded with a spinning rate of 12 kHz. Chemical shifts of ²⁷Al MAS NMR and ³¹P MAS NMR were referenced to (NH₄)Al(SO₄)₂ 12H₂O and diammonium phosphate at ‒0.4 ppm and 1.13 ppm, respectively.

The UV-vis spectrum was collected on Varian Cary 5000 UV-Vis-NIR spectrophotometer and BaSO₄ was used as a reference.

Electron paramagnetic resonance (EPR) spectrum was recorded on Bruker A 200 at 77 K to identify the isolated Cu²⁺. Before the test, samples were treated by N₂ at 285 K for 12 h. CuSO₄ solution was used to quantify the isolated Cu²⁺.

HT hydrothermal aging was performed at 973 K for 16 h in the presence of 10% H₂O in air.

LT hydrothermal aging was performed at 353 K for 24 h in the presence of 10% H₂O in N₂ atmosphere.

The NH₃-SCR activity was performed in a fixed bed with the gas hourly space velocity (GHSV) of 90, 000 h^‒1. 100 mg catalyst (60‒80 mesh) mixed with 200 mg quartz beads (60‒80 mesh) was transferred to a quartz tubular reactor. Prior to testing the activity, the catalyst was heated to 823 K (2 K/min) and maintained at this temperature for 30 min in the presence of feed gases. Then, the catalyst was cooled down gradually to 426 K and started to test. The feed gases are balanced by N₂ and the final component is 0.05% NO, 0.05% NH₃, 6.1% O₂, and 6.4% H₂O. Fourier transform infrared (FTIR) spectrometer (Tensor 27, Bruker) was used to detect NO, NO₂, and N₂O in the inlet and outlet gases.

The PXRD data (Fig. 2), scanning electron microscopy (SEM) images (Fig. S1), and BET results (Table S2) demonstrate the high crystallinity of SAPO-17. X-ray fluorescence spectroscopy (XRF) is used to determine the contents of Si, Al, P, and Cu, while the number of CHA and H₂O is obtained via thermo gravimetric analysis (TG) (Fig. S2). The obtained unit-cell chemical compositions and Cu contents are listed in Table 1 and S2. ¹H-²⁹Si CP MAS NMR spectrum (Fig. S3) indicates that all three samples contain silicon islands. Peaks at ‒20.2, ‒26.9, and ‒31.3 ppm in the ³¹P NMR spectrum (Fig. S4) can be assigned to different local environments of tetrahedral phosphorus in the framework [18]. The peak at 15.2 ppm in ²⁷Al NMR spectrum (Fig. S5) belongs to a portion of five-coordinated Al1 interacting with hydroxyl groups, while 35.2 and 43.8 ppm are assigned to tetrahedral Al2 and Al1, respectively [18]. The spectrum of ¹H-¹³C CP MAS NMR (Fig. 3) shows peaks at 26.94, 27.67, 34, and 56 ppm, indicating that the CHAs remain intact and protonated [18]. These results are also confirmed using in-situ DRIFT. As illustrated in Fig. 4, the bands at 3222, 1600, and 1502 cm^‒1 are assigned to the stretching and deformation vibration of NH₃⁺, indicating that CHAs are protonated [19]. It is notable that there is a sharp peak at 3629 cm^‒1, signifying the presence of extra framework OH^‒ [19], which is consistent with the ²⁷Al NMR result. Although these results indicate that CHAs are occluded as protonated form, their structural information, such as atomic coordinates and interactions with the framework remain unknown. Hence, we further investigate the locations of CHAs and host-guest interactions in the as-made SAPO-17 via Rietveld refinement against PXRD data.

Among the samples synthesized, we choose SAPO-17-8.0% as an example for further structural elucidation. The initial structure model of SAPO-17 was adopted from the International Zeolite Association (IZA). Si was treated and distributed evenly at four T sites. The adopted atomic coordinates of P, Al, and O are further optimized based on the chemical composition, and the experimental unit cell parameters are deduced from LeBail fitting. As the high angle data is not affected significantly by the guest species in the cages, we can search an appropriate scale factor between experimental and simulated data based on the data from 60° to 120° in 2θ. Applying this scale factor to the whole pattern, a difference Fourier map was obtained (Fig. S6). It is clear that each eri cage has two clouds of electron densities. Due to the characteristic features of CHA, a six membered ring and an NH₂ group, we can identify the initial positions of CHAs easily based on the difference Fourier map. The simulated annealing algorithm was then employed to determine the initial locations and orientations of CHAs (CHA is considered as a rigid body in this procedure). Finally, the atomic coordinates of P, Al, and O atoms, the zero shift, and the unit cell parameters, etc. are refined. The final Rietveld refinement was converged to R_p= 1.965%, R_wp = 2.707%, R_Bragg = 0.870%, and GOF = 1.699 (as shown in Table S3 and Fig. 5). Additionally, atomic coordinates, inter-atomic distances, and bond angles for the final refinement are provided in Table S4 and S5, respectively. The final refinement reveals that each eri cage accommodates two CHAs as deduced from TG (Fig. S2). The NH₃⁺ groups of the protonated CHAs point to the 8-ring pore openings and the classical hydrogen-bond N-H…O4 is identified, which indicates that, after the calcination, the proton bonded with O4 will be the Brönsted acid site. SAPO-17-8.0% has OH^‒ groups located in can cages and coordinated to Al1 atoms, as indicated by DRIFT and ²⁷Al NMR.

Inspired by past experiments regarding the Cu loading small-pore SAPO MSs for the NH₃-SCR reaction, we tentatively prepared Cu-SAPO-17 for this reaction. Fig. 6a shows the NO conversion of the fresh Cu-SAPO-17-2.7%-0.42, Cu-SAPO-17- 5.0%-0.47, and Cu-SAPO-17-8.0%-0.43. The activities of these three catalysts increase as follows: Cu-SAPO-17-2.7%-0.42 < Cu-SAPO-17-5.0%-0.47 < Cu-SAPO-17-8.0%-0.43. As shown in Fig. 7, the bands at 1440 and 1615 cm^‒1 are assigned to the N-H bending vibration of NH₃ adsorbed on Brönsted acid sites and Lewis acid sites, respectively. The bands at 3189, 3220, 3284, and 3357 cm^‒1 belong to NH₃ adsorbed on the Cu²⁺ ions, the asymmetrical and symmetrical vibration of NH₄⁺, and physically adsorbed NH₃, respectively [37]. The Brönsted acid sites of these three catalysts also increase as Cu-SAPO-17-2.7%-0.42 < Cu-SAPO-17-5.0%-0.47 < Cu-SAPO-17-8.0%-0.43. This is further confirmed by NH₃-TPD results (Figs. S7 and S8). In addition, the contents of Cu²⁺ also increase as Cu-SAPO-17- 2.7%-0.42 < Cu-SAPO-17-5.0%-0.47 < Cu-SAPO-17-8.0%-0.43 (Table 1). More Brönsted acid sites and Cu²⁺ were beneficial to NH₃-SCR activity [38, 39]. Thus, compared with the other two catalysts, Cu-SAPO-17-8.0%-0.43 shows the best NH₃-SCR activity.

Cu-SAPO-17-8.0%-0.43 displays the best activity among these three catalysts. Hence, we exchanged different amounts of Cu²⁺ to SAPO-17-8.0% and tested the influence of Cu contents over the NH₃-SCR. As illustrated in Fig. 6b, Cu-SAPO-17-8.0%-0.22 shows the best NO conversion among these three samples. The decrease of Cu²⁺ contents in Cu-SAPO-17-8.0%-0.11 (Table 1) results in its lower activity in the low-temperature region. Compared with Cu-SAPO-17- 8.0%-0.22, the activity of Cu-SAPO-17-8.0%-0.43 decreases in the high-temperature region, which might be attributed to the decrease of its crystallinity framework (Table S6) and a higher Cu content [40]. As mentioned earlier, the HT and LT hydrothermal stabilities of NH₃-SCR catalysts are important to practical applications. Moreover, we also investigated the activities of catalysts aged at 353 K under 10% H₂O/N₂ for 24 h and 973 K under 10% H₂O/air for 16 h to test their LT and HT hydrothermal stabilities, respectively. After HT hydrothermal aging, Cu-SAPO-17-8.0%-0.43 lost most of its activity (Fig. 6c). This is due to the collapse of its crystallinity framework as shown in Fig. S9 and Table S6. After HT hydrothermal aging, Cu-SAPO-17-8.0%-0.22 and Cu-SAPO-17-8.0%-0.11 showed comparable activities and they maintained most of the crystallinity framework (Figs. S10, S11, and Table S6). Considering the fresh and HT hydrothermal aging activity results, Cu-SAPO-17-8.0%-0.22 exhibits the best activity. Thus, when we tested its LT hydrothermal stability as shown in Fig. 6d, it maintains over 90% fresh activity in the low-temperature range (< 523 K) and nearly 100% fresh activity in the high-temperature range (> 523 K).

In order to further understand the properties of Cu-SAPO-17, UV-vis, and EPR were carried out. As shown in Fig. S12, fresh Cu-SAPO-17 catalysts display similar UV-vis spectrum, and they contain not only isolated Cu²⁺ but also Cu cluster and CuO_x. Therefore, we quantified isolated Cu²⁺ by EPR results (Table 1). Cu-SAPO-17 shows one signal of g_// = 2.355, A_// = 144 G (Fig. S13). Unfortunately, we cannot identify the locations of Cu²⁺ through EPR, only four suitable sites were speculated in the previous literature [17]. We failed to locate the Cu²⁺ through refining the PXRD data of Cu-SAPO-17 sample, due to the low Cu²⁺ contents. Fortunately, we have identified the Brönsted acid sites (bonded with O4) through refining the PXRD data of the as-made SAPO-17. Al and P atoms distribute alternately in AlPO molecular sieves, thus, ignoring the influence of Si island; Si atoms mainly substitute P1 in SAPO-17-8.0%. Considering orientations of O atoms and the charge balance in Cu-SAPO-17, we can deduce the possible locations of Cu²⁺: site I, adjacent to the 6-ring window joining the can cage and d6r but displaced into can cages; site II, adjacent to the s6r joining the can and eri cage but displaced into eri cage (Fig. 8). The accurate locations of Cu²⁺ in Cu-SAPO-17 are still under investigation.

In conclusion, SAPO-17 was hydrothermally synthesized by commercialized cyclohexylamine and their Si contents can be tunable. Rietveld refinement results reveal that CHAs are located in the ellipsoidal eri cages and H atoms of NH₃⁺ groups formed the hydrogen bond with O4. Finally, SAPO-17 samples were exchanged with Cu in the aqueous solution and tested their NH₃-SCR activities. Cu-SAPO-17-8.0%-0.22 exhibits promising activities even after being aged at 353 K for 24 h or 973 K for 16 h. The methodology regarding the structural elucidation by Rietveld refinement can be utilized in other crystalline porous materials in the near future.