Dynamic evolution of HZSM-5 zeolite framework under steam treatment

ZSM-5 zeolite preparation and steam treatment

The ZSM-5 (Si/Al = 36) zeolite was synthesized following the procedure described in the literature^[52]. A detailed synthesis description is provided in the Supplementary Material. To obtain HZSM-5 zeolites, calcination at 550 °C for six hours was performed to remove the organic template. The detailed steam treatment of HZSM-5 zeolites followed the approach previously developed by our group^[35]. The schematic diagram of high-temperature and high-pressure steam treatment is shown in Supplementary Figure 1. Prior to steam treatment, the HZSM-5 zeolite was dehydrated under vacuum (< 10^-3 Pa) at 420 °C for over 12 h in a dry quartz tube, and then, it was transferred into a glovebox under a dry argon (Ar) atmosphere for subsequent operations. 10ul deionized water and 50 mg dry HZSM-5 zeolites were added into a quartz tube, separated by a layer of quartz wool. Finally, the quartz tube was placed into a stainless autoclave to seal, and the autoclave was further placed in a muffle furnace and heated at different temperatures (T = 100, 200, 300, 400, and 500 °C) for a period of times. The treated samples were named HZSM-5-T-X (T and X represent the heating temperature in degrees Celsius and heating times in hours, respectively). Note that the untreated HZSM-5 zeolite was named as HZSM-5. Furthermore, the steamed samples were converted into the ammonium form through ammonium ion exchange, denoted as HZSM-5-T-X-NH₄, as detailed in the Supplementary Material.

Characterization

The powder X-ray diffraction (XRD) patterns were recorded on a PANalytical X’Pert PRO X-ray diffractometer with Cu Kα radiation (λ = 1.5418 Å, 40 kV, 40 mA) to confirm the MFI-type topology structure of the samples. The chemical compositions of the samples were measured on an X-ray fluorescence (XRF) spectrometer (Philips Magix-601). The crystal morphology and particle size of the samples were characterized by field emission scanning electron microscopy (FE-SEM, Hitachi, TM3000). The N₂ adsorption-desorption experiments were performed to obtain more information about the pore structure at 77 K (Micromeritics ASAP 2020). The temperature-programmed desorption of ammonia (NH₃-TPD) was carried out with a Micromeritics Autochem II 2920 analyzer. Subsequently, 150 mg samples were outgassed at 550 °C for 1 h under helium flow, then cooled to 100 °C and exposed to NH₃ to saturate the adsorption. The desorption process was carried out from 100 to 800 °C. Fourier transform infrared (FTIR) spectra were recorded on a Nicolet IS50 FTIR spectrometer. The samples (about 5 mg) were pressed into a self-supporting wafer and placed into a quartz cell. Before the measurement, the samples were pretreated at 300 °C for 1 h under vacuum, and then, the spectrum was recorded immediately after the pretreatment.

¹H, ²⁷Al, and ²⁹Si MAS NMR and ¹H-²⁹Si cross-polarization (CP) MAS NMR experiments were performed on a Bruker Avance NEO 500 spectrometer equipped with 11.7 T and 89 mm wide-bore magnet using a 3.2 mm H/F-X-Y triple resonances MAS probe with the corresponding Larmor frequency of 500.13 MHz, 130.32 MHz, and 99.36 MHz for ¹H, ²⁷Al, and ²⁹Si, respectively. Prior to all ¹H MAS NMR, the samples were dehydrated under vacuum (< 10^-3 Pa) at 420 °C for over 12 h in a dry quartz tube connected to a vacuum line and then were transferred into 3.2 mm NMR rotors (tightly sealed by a Kel-F cap) under a dry Ar atmosphere in a glovebox. The chemical shifts of ¹H NMR were referenced to adamantane at 1.74 ppm. ¹H MAS NMR spectra were performed with a π/2 pulse width of 3.8 μs, a recycle delay of 70 s, and a spinning rate of 20 kHz. The chemical shift of ²⁷Al NMR was referenced to 1 mol/L Al(NO₃)₃ solution at 0 ppm. ²⁷Al MAS NMR spectra were performed with a pulse width of 0.63 μs (π/18), a recycle delay of 0.5 s, and a spinning rate of 20 kHz. The chemical shifts of ²⁹Si NMR were referenced to kaolinite at -91.5 ppm. ²⁹Si MAS NMR spectra were performed with high-power proton decoupling using a π/4 pulse width of 2.5 μs, a recycle delay of 10 s, and a spinning rate of 10 kHz. ¹H-²⁹Si CP MAS NMR spectra were performed with a contact time of 3 ms, a recycle time of 2 s, and a spinning rate of 10 kHz. The decoupling field of 69.4 kHz was applied during the acquisition time.

We conducted several experiments on 2D ¹H-¹H DQ-SQ MAS NMR, ¹H-²⁷Al symmetry-based rotational-echo saturation-pulse double-resonance (S-RESPDOR) MAS NMR, and ²⁷Al MQ MAS NMR on a Bruker Avance III 600 spectrometer equipped with 14.1 T and 89 mm wide-bore magnet using a 3.2 mm H-X-Y triple resonance MAS probe with the corresponding Larmor frequencies of 600.13 MHz and 156.38 MHz for ¹H and ²⁷Al, respectively. For the 2D ¹H-¹H DQ-SQ MAS NMR experiments, we employed the POST-C7 pulse sequence, accumulating either 128 or 400 scans for each of the 64 points in t1 dimensions with an increment time of 41.67 μs, a recycle delay of 2 s, and a spinning rate of 12 kHz. In the case of ¹H-²⁷Al S-RESPDOR spectra, we implemented a recycle delay of 10 s and a spinning rate of 20 kHz. For the ¹H channel, we applied a π/2 pulse width of 3.2 μs and a π pulse width of 6.4 μs using SR4 recoupling. The recoupling time ranged from 0.2 to 1.0 ms, with intervals of 0.2 ms. The saturation pulse on the ²⁷Al channel was with a duration of 75 μs (1.5 Tr). ²⁷Al MQ MAS NMR spectra were obtained using the z-filter 3QMAS pulse sequence. The pulse durations were set to 4.2 μs and 1.4 μs for the first and second pulse, and the pulse width of the selective pulse was set to 15 μs. Additionally, 3000 or 15000 scans were accumulated for each of the 64 points in t1 dimensions with an increment time of 50 μs, a recycle delay of 0.1 s, and a spinning rate of 20 kHz. The pulse sequences are shown in Supplementary Figure 2A-C.

We conducted 2D ²⁷Al-²⁷Al DQ-SQ MAS NMR experiments on a Bruker Avance III 600 spectrometer equipped with 14.1 T and 89 mm wide-bore magnet using a 4.0 mm H-X double resonance MAS probe. DQ coherences were excited and reconverted using the BR2¹₂ pulse sequence with τ_exc = τ_rec = 571.43 μs. CT-selective π/2 and π pulse widths of 11 μs and 22 μs, a recycle delay of 0.1 s, and a spinning rate of 14 kHz were used for the ²⁷Al-²⁷Al DQ-SQ MAS NMR experiments. Furthermore, 40,960, 64,000,120,000, and 128,000 scans were accumulated for each of the 32 points in t1 dimensions with an increment time of 35.71 us for the HZSM-5, HZSM-5-300-24, HZSM-5-400-2, and HZSM-5-500-2 samples, respectively. The pulse sequence is shown in Supplementary Figure 2D.

Structural transformation of HZSM-5 zeolite before and after steam treatment

A series of untreated and steam-treated HZSM-5 samples are characterized by XRD and scanning electron microscopy (SEM). The similar XRD patterns [Supplementary Figure 3A] of the HZSM-5-T-X samples indicate that the MFI-type topology is still maintained after high-temperature and high-pressure hydrothermal treatment. However, as the temperature rises to 200 °C, the peak at 24.4 °C split into a double peak [Supplementary Figure 3B], representing clear evidence for transition of the MFI-type structure from orthorhombic to monoclinic symmetry with 12 and 24 distinguishable framework T sites, respectively^[53-55]. This phenomenon can be explained by a mutual shift of successive (010) pentasil layers along [001], which is closely related to the distortion or rotation of the T₄ and T₆ rings interconnecting the pentasil layers^[56]. Typically, this transformation is influenced by the Si/Al ratio in the lattice or the adsorption of molecules within ZSM-5 and can be measured by ²⁹Si MAS NMR^[57,58]. In addition, taking HZSM-5 as a reference, the relative crystallinity does not start to decrease until the hydrothermal temperature rises to 400 °C, with the lowest value being 87% at 500 °C [Table 1], indicating that harsh hydrothermal treatment can cause the degradation of the MFI-type framework. The morphology and size of particles exhibit no significant changes before and after steam treatment, as can be seen from SEM images [Supplementary Figure 4]; yet, this does not necessarily imply the preservation of zeolite framework microstructure at the atomic level.

²⁷Al MAS NMR is a potential tool for tracking the dynamic evolution of aluminum species during hydrothermal dealumination^[29,59]. ²⁷Al single-pulse MAS NMR spectra of the pristine and steam-treated HZSM-5 samples are recorded, as shown in Figure 1A. For HZSM-5 samples, a solitary signal at 55 ppm is evident, representing the framework tetrahedral-coordinated Al species, denoted as Al(IV)-1^[43]. When the steam temperature is below 200 °C, the main signal at 55 ppm remains nearly constant in intensity. However, upon reaching 300 °C, the intensity of Al(IV)-1 slightly decreases, accompanied by the emergence of a weak broad signal at 45 ppm. Following treatment at 400 or 500 °C, the intensity of Al(IV)-1 significantly diminishes, and two distinct signals gradually emerge at 30 ppm and 0 ppm, increasing in intensity as Al(IV)-1 decreases. The signals at 30 ppm and 0 ppm are assigned to the penta-coordinated Al and the octahedral-coordinated Al, respectively. Both the penta-coordinated and octahedral-coordinated Al species remain apparent, with minimal changes in relative content, after back exchanging to their ammonium form [Supplementary Figure 5], implying that they can be considered as EFAL species rather than framework-associated Al^[37]. Nevertheless, note that ²⁷Al is a nucleus with S = 5/2, and its resolution is limited due to the residual second-order quadrupole broadening that is not entirely eliminated by magic-angle spinning. Therefore, in order to qualitatively identify the Al species in different chemical environments better, the combination of ²⁷Al MQ MAS NMR is indispensable^[43,60,61]. The MQ MAS NMR method, proposed by Frydman and Harwood, is capable of refocusing second-order quadrupole effects, thereby obtaining the isotropic high-resolution spectrum of half-integer quadrupole nuclei, such as ¹⁷O, ²³Na, and ²⁷Al^[62]. NMR parameters of different Al species extracted from ²⁷Al MQ MAS NMR spectra are listed in Supplementary Table 1. In the ²⁷Al MQ MAS NMR spectra of HZSM-5 [Figure 1B] and HZSM-5-200-2 [Figure 1C], only the Al(IV)-1 signal remains present, illustrating that dealumination does not occur under mild steam conditions. When the temperature increases to 300 °C, as shown in Figure 1D and Supplementary Figure 6, a new broad signal emerges around 45 ppm, exhibiting a larger quadrupole coupling constant (C_Q = 5.90 MHz), attributed to the partially bonded framework (SiO)_4-n-Al(OH)_n species, denoted as Al(IV)-2, demonstrating the irreversible hydrolysis of the first framework Al-O bond is achieved.^[42] Prolonging the steam treatment time at 300 °C to 24 h [Supplementary Figure 5], the content of Al(IV)-2 increases and EFAL at 0 and 30 ppm appears, signifying that Al(IV)-2 is a critical intermediate species for the complete dealumination of the framework. Surprisingly, Al(IV)-2 persists distinctly, with minimal changes in relative content, even after exchanging HZSM-5-300 into ammonium form, as shown in Supplementary Figure 5 and Figure 1E. This implies that the transformation from Al(IV)-2 to Al(IV)-1 is not achievable through ammonium exchange treatment. This phenomenon differs from the reversible octahedral–tetrahedral coordination transformation observed in the adsorption of water framework-associated Al^[63]. After treatment at 400 or 500 °C, the same phenomenon is observed [Figure 1A and F], with the extent of dealumination escalating as the hydrothermal conditions become more severe. Hence, both hydrothermal temperature and time are the primary factors affecting the stability of the zeolite framework. It is not difficult to obtain the dynamic evolution path of the framework tetrahedrally coordinated Al atom in severe hydrothermal environments: the four framework Al-O bonds bonded to a framework Al atom partially hydrolyze to form Al(IV)-2 first, and subsequently, the Al atom can be removed from the framework to generate EFAL until all the four framework Al-O bonds are completely hydrolyzed by water molecules.

Figure 1. (A) ²⁷Al single-pulse MAS NMR spectra of HZSM-5 before and after steam treatment at 100 ~ 500 °C for 2 h. Note that the intensities have been normalized by mass. ²⁷Al MQ MAS NMR spectrum of HZSM-5 (B); HZSM-5-200-2 (C); HZSM-5-300-2 (D); HZSM-5-300-2-NH₄ (E) and HZSM-5-500-2 (F). All samples are fully hydrated.

To gain further insight into the local structural environments surrounding Si atoms, ²⁹Si MAS NMR [Figure 2A] is adopted. Five primary peaks at -103, -106.5, -110, -113, and -116 ppm are observed in all the samples from low to high fields. The signal at -103 ppm can be attributed to Q³ [Si(OSi)₃(OH)] species^[29,64], whose intensity can be obviously enhanced in ¹H-²⁹Si CP MAS NMR spectra [Supplementary Figure 7] through the transfer of magnetization from dipolar-coupled ¹H to ²⁹Si. Correspondingly, the peak at -106.5 ppm is ascribed to Si1Al (Si(OSi)₃(OAl)) species^[29,64], and its relative percentage is nearly unchanged after steaming at 100 °C and 200 °C for two hours [Figure 2B and Supplementary Table 2]. In other words, there is a slight change in the framework Si/Al ratio, which is about 37, demonstrating that mild steam conditions do not cause irreversible hydrolysis of the framework. In contrast, its relative percentage begins to decline from 10.78% to 2.96% after severe steam treatment at temperatures ranging from 300 °C to 500 °C [Figure 2B and Supplementary Table 2], accompanied by a substantial increase in the framework Si/Al ratio from 37 to 135, suggesting that harsh steam conditions can cause the irreversible hydrolysis of the framework Si-O-Al bonds. The peaks at -110 ppm, -113 ppm, and -116 ppm are all assigned to Q⁴ (Si(OSi)₄) species^[29,64], and their total relative percentage increases from 88.56% to 96.88% as the temperature rises to 500 °C [Figure 2B and Supplementary Table 2], demonstrating dealumination and healing of defects sites occur simultaneously. It is noteworthy that the resolution of ²⁹Si MAS NMR spectra is enhanced after steaming at 400 °C and 500 °C, which can be ascribed to the transition from the orthorhombic to monoclinic symmetries, the reduction of defect Si sites, and the improvement of Si/Al by dealumination procedures^[57,65]. The presence of defect Si sites and Al atoms in zeolites can induce a distribution of Si-O bonds and Si-O-Si angles, which broaden the spectral lines and obscure the detection of various crystallographically distinct T sites. These findings are consistent with the results obtained from XRD [Supplementary Figure 1] and ²⁷Al MAS NMR [Figure 1] mentioned above.

Figure 2. (A) ²⁹Si MAS NMR spectra of HZSM-5 before and after steam treatment; (B) relative percentage of Si species in different chemical environments by deconvolution of ²⁹Si MAS NMR spectra.

In order to identify the dynamic evolution of hydroxyl groups, ¹H MAS NMR spectroscopy and FTIR are employed. ¹H MAS NMR spectra are recorded [Figure 3A] to quantify the concentrations of various hydroxyl groups by deconvolution. The signals at 3.9, 2.0, and 1.7 ppm are assigned to the unperturbed BAS, geminal or vicinal Si-OH groups, and isolated Si-OH groups, respectively^[66,67]. Generally, the signal at 2.5 ppm is attributed to Al-OH groups that exist extra-lattice or EFAL, most likely as small aluminum oxide moieties^[67]. Notably, this signal has been verified as framework Al-OH groups on Al(IV)-2 in HZSM-5 zeolites but not EFAL in recent studies^[42,43]. Additionally, this signal can also be assigned to water molecules belonging to Al(OH)_n(H₂O) species from EFALs or external surface sites and hydrogen-bonded silanols^[68-71]. Therefore, the precise assignment of this signal remains controversial. In view of this, its attribution is discussed below. Furthermore, the assignment of the broad shoulder peak at 4.6 ppm is relatively complex and remains controversial. So far, it has been assigned to hydrogen-bonded BAS^[72,73], residual water^[74], and hydrogen-bonded silanols^[70] in previous studies. Consequently, a more detailed discussion regarding its assignment is provided below. The broad signal at 12.0-15.0 ppm can be assigned to the BAS on Al(IV)-2^[42,43]. Obviously, the concentration of the total hydroxyl groups gradually decreases from 0.498 mmol/g to 0.105 mmol/g as the temperature increases to 500 °C [Figure 3B and Supplementary Table 3], indicating the occurrence of hydroxyl condensation. After mild steam treatment (T ≤ 200 °C), the concentration of BAS at 3.9 ppm remains essentially stable at around 0.1 mmol/g. However, interestingly, the concentration of the broad shoulder peak at 4.6 ppm decreases significantly from 0.262 mmol/g to 0.146 mmol/g. Considering the framework Si/Al ratio remains relatively constant, as indicated by ²⁹Si NMR [Figure 2], and no other Al species besides Al(IV)-1 are observed in ²⁷Al NMR [Figure 1A-C], this situation mentioned above can be attributed to the condensation of numerous silicon hydroxyl groups that are inevitably generated during synthesis and calcination. This is further corroborated by the IR results below. Upon reaching 300 °C, the concentrations of Al-OH groups at 2.5 ppm and another broad signal at 12.0-15.0 ppm reach the maximum, but only Al(IV)-2 is generated from ²⁷Al NMR [Figure 1A and D], indicating the signal at 2.5 ppm is related to Al(IV)-2. Meanwhile, the internuclear ¹H-²⁷Al distance of this signal is 2.05 ± 0.11 Å by the ¹H-²⁷Al S-RESPDOR experiments [Supplementary Figure 8], indicating that the proton is closely bound to Al atoms. Therefore, it is reasonable to attribute this signal to the framework Al-OH group on Al(IV)-2, consistent with recent studies^[42,43,68]. Severe steam treatment (T ≥ 300 °C) can induce framework dealumination, which has been demonstrated by ²⁷Al and ²⁹Si MAS NMR [Figures 1 and 2]. Strangely, only the signal at 2.5 ppm is observed in ¹H NMR of dealuminated HZSM-5 zeolites, with no additional new signals emerging. This prompts the consideration that Al-OH groups on EFAL may also contribute to the signal at 2.5 ppm. This possibility cannot be ruled out entirely. That is, the signal at 2.5 ppm cannot be simply attributed to Al-OH groups on EFAL or Al(IV)-2 species and requires thorough analysis alongside ²⁷Al and ²⁹Si NMR. Undeniably, the enhancement of the signal at 2.5 ppm is indicative of the irreversible hydrolysis of the Si-O-Al bonds. In principle, the complete removal of a framework Al atom should result in an Al(OH)₃·H₂O group and a silicon nest consisting of four silicon hydroxyl groups with hydrogen bonding interactions. Unexpectedly, the concentrations of Al-OH groups and Si-OH groups significantly decrease, which can occur through the condensation reaction from the hydroxyl groups bonded on Al atoms, forming new EFAL species with fewer hydroxyl groups, and from the expected four Si-OH groups (silanol nest) forming new framework Si-O-Si bonds, respectively^[75]. The condensation of Al-OH groups causes the diversity of EFAL, such as Al(OH)₃, Al(OH)₂⁺, Al(OH)²⁺, AlO⁺, Al₂O₃, dimers, trimers, etc. The formation of Si-O-Si bonds may be caused by the direct condensation of adjacent Si-OH groups in the silanol nest and the migration of Si(OH)₄ species formed by desilication to the silanol nest for reorganization. The former may be under mild hydrothermal conditions, where desilication is improbable. However, severe hydrothermal conditions may cause the removal of Si atoms from the framework, making the latter possibility plausible.

Figure 3. (A) ¹H MAS NMR spectra of HZSM-5 before and after steam treatment; (B) concentrations of different hydroxyl groups by quantifying and fitting the quantitative ¹H MAS NMR spectra; (C) FTIR spectra of the OH vibration region of HZSM-5 before and after steam treatment; (D) ¹H-²⁷Al S-RESPDOR spectra of HZSM-5-300-2 sample with a recoupling time of 0.6 ms. S and S’ denote the ¹H spectrum without and with ²⁷Al irradiation, respectively; ΔS is the difference spectrum obtained by S-S’.

Infrared (IR) spectra are also utilized to further investigate the dynamic evolution of hydroxyl groups, as shown in [Figure 3C]. In this spectra, five main IR bands are observed in the region of 3,800-3,400 cm^-1. The bands at 3,743 cm^-1, 3,725 cm^-1, 3,608 cm^-1, and 3,500 cm^-1 are ascribed to external Si-OH groups, internal Si-OH groups, BASs, i.e., Si(OH)Al groups, and silanol nests, respectively^[76]. The band at 3660 cm^-1 can be assigned to Al-OH groups on EFAL^[77,78] or partially framework-bonded Al species^[79,80], but this remains controversial. It can be found that mild steam treatment (T ≤ 200 °C) does not decrease the intensity at 3,743 cm^-1, 3,660 cm^-1, and 3,608 cm^-1 but significantly decreases the intensity at 3,725 cm^-1 and 3,500 cm^-1, which is indicative of selective removal of Si-OH groups. This observation is consistent with ¹H MAS NMR results and provides further evidence that the silanol nests, hydrogen-bonded silanols, contribute to a portion of the broad shoulder peak at 4.6 ppm in ¹H MAS NMR spectra. After steam treatment at 300 °C, the intensity of the band at 3660 cm^-1 remarkably increases. Given that only Al(IV)-2 is generated from ²⁷Al NMR [Figure 1A and D], the band at 3,660 cm^-1 can be attributed to Al-OH groups on Al(IV)-2 and is associated with the signal at 2.5 ppm in ¹H NMR. The lower intensity of the band at 3608 cm^-1 for severe steam conditions (T ≥ 300 °C) suggests that less framework tetrahedral Al compared to HZSM-5, in agreement with the lower Al(IV)-1 in ²⁷Al MAS NMR [Figure 1A] and the higher framework Si/Al ratio in ²⁹Si MAS NMR [Figure 2B]. Similarly, the intensity of the band at 3660 cm^-1 gradually decreases. It is worth mentioning that no additional vibrational bands belonging to Al-OH groups on EFAL species are observed in dealuminated HZSM-5 zeolites, which is a puzzling observation. Therefore, it is reasonable to speculate that the band at 3,660 cm^-1 represents the superposition of Al-OH groups on Al(IV)-2 and EFAL species.

The acidic properties are examined through NH₃-TPD experiments, as shown in Supplementary Figure 9. The two ammonium-desorption peaks at around 470 and 680 K are typically attributed to NH₃ desorption from weak and strong acid sites, respectively^[76]. Evidently, the temperature and intensity of these two ammonium-desorption peaks remain nearly unchanged after mild steam treatment (T ≤ 200 °C) but gradually decrease as the temperature increases (T ≥ 300 °C). It can be inferred that the weakening of acid strength is a consequence of dealumination, possibly linked to changes in the local chemical environment.

The changes in textural properties induced by steaming are studied by N₂ adsorption experiments [Supplementary Figure 10A]. The isotherms of HZSM-5 steamed at no higher than 300 °C can be considered as a combination of type I and type IV isotherms, while the isotherms of HZSM-5 steamed at 400 and 500 °C can be regarded as type V isotherms. The textural properties of these samples are summarized in Table 1. As expected, the micropore volume remains essentially unchanged after mild steam treatment (T ≤ 200 °C); however, it obviously decreases after severe steam treatment (T ≥ 300 °C), which can be explained by EFAL blocking the channels. Interestingly, despite a more pronounced degree of dealumination at 500 °C compared to 400 °C, the micropore volume and surface slightly increase and the pore diameter is smaller [Table 1 and Supplementary Figure 10B], which may be related to the restructuring of amorphous Si species generated by desiliconation.

Identification of spatial proximity of various hydroxyl species

In order to determine the spatial proximity (or interaction) between different hydroxyls and Al species, the ¹H-²⁷Al S-RESPDOR experiment is conducted on HZSM-5-300-2, as shown in Figure 3D. Under ²⁷Al irradiation, the ¹H signal of protons that are in close proximity to ²⁷Al atoms will be modulated by ¹H-²⁷Al dipolar interaction. Only the protons observed in the difference spectrum (ΔS), acquired by subtracting the ¹H MAS NMR spectrum with ²⁷Al irradiation (S’) from that without ²⁷Al irradiation (S), have a dipolar interaction with Al atoms. In the difference spectrum [Figure 3D], the ¹H signals from BAS at 3.9 ppm, Al-OH at 2.5 ppm, three newly observed signals at 2.3 ppm, 2.8 ppm, and 3.2 ppm, along with three broad signals at 4.6 ppm, 6.0-9.0 ppm, and 12.0-15.0 ppm, are all discernible due to direct ¹H-²⁷Al dipole-dipole interactions; that is to say, the protons of these signals are close to the Al species. In contrast, there is no ¹H signal from Si-OH at 1.7 ppm in the difference spectrum. Note that the broad shoulder peak at 4.6 ppm with direct ¹H-²⁷Al dipole-dipole interaction can be assigned to hydrogen-bonded BAS^[73]. Combining the quantitative results of ¹H MAS NMR spectra and IR data, this broad signal at 4.6 ppm observed in ¹H MAS NMR spectra is at least produced by the superposition of hydrogen-bonded silanols and BAS, in line with previous studies^[70,72,73]. Besides, the three signals at 2.3, 2.8, and 3.2 ppm can be tentatively assigned to Al-OH groups, but their detailed structures are still unclear due to the diversity of Al-OH groups in zeolites.

To gain a deeper insight into the dynamic evolution and spatial proximity (or interaction) of different hydroxyl groups within the HZSM-5 zeolite framework during the hydrothermal process under high-temperature and high-pressure conditions, 2D ¹H-¹H DQ MAS NMR is adopted. This technique has been previously proven effective for obtaining spatial proximity information between various hydroxyl groups with proton-proton distances less than 5Å in a variety of zeolites, such as HZSM-5, HSSZ-13, HMOR, HY, etc.^[28,35,40]. Peaks that appear along the diagonal represent the autocorrelation peaks (ω, 2ω) resulting from the dipolar interaction of the proton species with the same chemical environment, while paired peaks that appear off-diagonal at (ω_a, ω_a + ω_b) and (ω_b, ω_a + ω_b) represent the correlations between proton species with different chemical shifts. Figure 4 shows the 2D ¹H-¹H DQ-SQ MAS NMR spectra of HZSM-5 before and after steaming. For clarity, all ¹H signals appearing from the high to low field are relabeled with letters from A to H, corresponding to the following chemical shifts: 1.7 ppm (A), 2.3 ppm (B), 2.5 ppm (C), 2.8 ppm (D), 3.2 ppm (E), 3.7 ppm (F), 6.0-9.0 ppm (G), and 12.0-15.0 ppm (H). For HZSM-5, two autocorrelation peaks (A-A and F-F) and one off-diagonal peak pair (C-H) are clearly observed in Figure 4A. The one autocorrelation peak (A-A) appearing at (1.7, 3.4) ppm corresponds to the spatial proximity between nonacidic silanol groups. Another autocorrelation peak (E-E) at (3.7, 7.4) ppm suggests that the bridge hydroxyl groups (Si-O(H)-Al) are not isolated but with spatial proximity, which is further supported by the following 2D ²⁷Al-²⁷Al DQ-SQ MAS NMR experiments [Figure 5]. This phenomenon can be explained by the non-homogeneous distribution of framework Al, considering that each unit cell has only 2.6 Al atoms calculated from the Si/Al of 36. The one off-diagonal peak pair (C-H) at (2.5, 14.5-17.0) and (12.0-14.5, 14.5-17.0) ppm is attributed to the correlation between framework Al-OH groups and BAS on Al(IV)-2 species that is inevitably generated during the calcination process for template removal. Compared to HZSM-5, after steaming at 100 °C and 200 °C, as shown in Figure 4B and C, no new autocorrelation and off-diagonal peak pair appear. Note that the green-doted region associated with the broad shoulder peak at 4.5-8 ppm disappears at 200 °C. This change may be the result of condensation of extensive hydrogen-bonded silanols, as corroborated by ¹H MAS NMR and FTIR [Figure 3A-C] above.

Figure 4. 2D ¹H-¹H DQ-SQ MAS NMR spectra of dehydrated HZSM-5 (A); HZSM-5-100-2 (B); HZSM-5-200-2 (C); HZSM-5-300-2 (D); HZSM-5-400-2 (E); HZSM-5-500-2 (F); (G-J) structure of proton signals with different ¹H chemical shifts where the letters from A to H corresponded to eight proton signals from the high filed to the low field, respectively.

Figure 5. 2D ²⁷Al-²⁷Al DQ-SQ MAS NMR spectra of hydrated HZSM-5 (A); HZSM-5-300-24 (B); HZSM-5-400-2 (C) and HZSM-5-500-2 (D). All samples were fully hydrated.

After steaming at 300 °C, the spectrum [Figure 4D] exhibits increased complexity. Apart from the four signals observed above, four new signals at approximately 2.3 ppm (B), 2.8 ppm (D), 3.2 ppm (E), and 6.0-9.0 ppm (G) are visible, which result in one weak autocorrelation peak and seven off-diagonal peak pairs being additionally observed. ²⁷Al MAS NMR spectra [Figure 1] have indicated that the framework Al-O bonds have been hydrolyzed to generate Al(IV)-2 species. Therefore, it is reasonably speculated that the four new signals (B, D, E, and G) are associated with Al(IV)-2 species. The off-diagonal peak pair (C-F) at (2.5, 6.2) and (3.7, 6.2) ppm can correspond to the correlation between Al-OH groups on Al(IV)-2 and unperturbed BAS. For broad signal at 6.0-9.0 ppm (G) adjacent to Al atoms in ¹H-²⁷Al S-RESPDOR [Figure 3D], this signal was detected in dehydrated HZSM-5 zeolites by ¹H-²⁷Al D-HMQC experiments and assigned to partially coordinated framework Al, Al(IV)-2^[43]. An analogous broad signal appearing at 5.0-8.0 ppm in the ¹H MAS NMR spectrum of CHA-type zeolites is also observed by our group. This signal has been attributed to BAS perturbed by framework Al-OH^[35]. In this work, the broad signal at 6.0-9.0 ppm (G) is assigned to the BAS on Al(IV)-2. Since the detailed structures of Al(IV)-2 are not single, it is reasonable to ascribe both broad signals at 6.0-9.0 ppm and 12.0-15.0 ppm to the BAS on Al(IV)-2. This can be supported by differences in spatial proximity to different protons in 2D ¹H-¹H DQ-SQ MAS NMR spectra [Figure 4D]. Considering that the BAS on Al(IV)-2 can be perturbed by framework Al-OH groups on Al(IV)-2 due to complex hydrogen bonding interactions, the broad signal at 6.0-9.0 ppm and 12.0-15.0 ppm may contain fewer and more Al-OH groups, respectively, as shown in Figure 4H and I. Correspondingly, the weak off-diagonal peak pair (F-G) at (3.7, 11.0) and (7.3, 11.0) ppm is attributed to the correlation between unperturbed BAS and hydrogen-bonded BAS of Al(IV)-2, which can be explained by partial hydrolysis of one of the two unperturbed BASs with spatial proximity. The new weak autocorrelation peak (G-G) at (7.0-8.0, 14.0-16.0) ppm corresponds to the spatial proximity of the Al(IV)-2, suggesting that small amounts of unperturbed BASs with spatial proximity are partially hydrolyzed simultaneously. Note that the broad signal at 6.0-9.0 ppm (G) is also proximate to the Al-OH groups at 2.3 ppm (B-G); therefore, it is speculated that the signal at 2.3 ppm (B) is also framework Al-OH groups in Al(IV)-2. Similarly, the four signals at 2.3 ppm (B), 2.5 ppm (C), 2.8 ppm (D), and 3.2 ppm (E) with spatial proximity to the broad signal at 12.0-15.0 ppm can also be attributed to framework Al-OH groups in Al(IV)-2. Importantly, unlike the signals at 2.3 ppm (B) and 2.5 ppm (C), the new Al-OH groups at 2.8 ppm (D) and 3.2 ppm (E) only have spatial proximity with the broad signal at 12.0-15.0 ppm (D-H and E-H), without another broad signal at 6.0-9.0 ppm (G), which also points more complex interactions between the signal at 12.0-15.0 ppm (H) and Al-OH groups. The diversity of Al-OH groups on Al(IV)-2 may be caused by the complex and variable chemical environment.

Moreover, the spectrum of HZSM-5-400-2 [Figure 4E] closely resembles that of HZSM-5-300-2 [Figure 4D]. Two subtle distinctions are noteworthy. Firstly, the off-diagonal peak pair (F-G) at (3.7, 11.0) and (7.3, 11.0) ppm disappears. Secondly, the autocorrelation peaks (F-F and G-G) at (3.7, 7.4) and (8.0, 16.0) ppm weaken and strengthen, respectively. These indicate that more Al(IV)-1 is partially hydrolyzed to form Al(IV)-2; that is, increasing the temperature can accelerate the irreversible hydrolysis process of framework Al-O bonds, aligning with the framework dealumination observed in ²⁷Al MAS NMR [Figure 1]. The initial Al(OH)₃·H₂O EFAL species, generated by removal of Al atoms from the framework, would evolve further by interacting with BAS within the zeolite channels^[27,28]. As analyzed above, the possibility that the signals at 2.3-3.2 ppm (B-E) come from EFAL species cannot be ignored, given the complexity of the structural and chemical environments of EFAL and Al(IV)-2 species, along with the limited resolution of ¹H NMR spectra. At 500 °C, many signals with spatial proximity vanish [Figure 4F], which is primarily because most of the Al(IV)-1 are completely hydrolyzed and eventually evolved into EFAL species with fewer hydroxyl groups through dehydration condensation. Obviously, during dealumination of HZSM-5 under steam conditions, the spatial proximities (or interactions) between various hydroxyl groups in structurally distinct Al species (Al(IV)-1, Al(IV)-2, and EFAL species) are exceptionally intricate.

Identification of spatial proximity of various aluminum species

To identify the spatial proximity (or interaction) among various aluminum species in both pristine and steamed HZSM-5 zeolites, a 2D ²⁷Al-²⁷Al DQ-SQ MAS NMR technique is applied, as shown in Figure 5. For untreated HZSM-5 zeolite [Figure 5A], only one autocorrelation peak at (56, 112) ppm is observable, indicating that the tetrahedrally coordinated framework Al species, Al(IV)-1, are in close proximity to each other, in accordance with the observation of autocorrelation peak of BASs in the 2D ¹H-¹H DQ-SQ MAS NMR spectrum [Figure 4A]. After steaming at 300 °C for 24 h, compared to Figure 5A, one additional autocorrelation peak and two off-diagonal peak pairs are clearly distinguished, as shown in Figure 5B. The emergence of autocorrelation at (2, 4) ppm indicates that six-coordinate EFAL species are in close proximity to each other. A cross-peak pair at (2, 57) and (55, 57) ppm is attributed to the correlation between six-coordinate EFAL species and Al(IV)-1, and another weaker cross-peak pair at (0, 31) and (31, 31) ppm corresponds to the correlation between six- and five-coordinate EFAL species. Yu et al. studied the dealumination mechanism of HY, HMOR, and HZSM-5 zeolites, combining ²⁷Al DQ MAS NMR and theoretical calculations, and proposed that it is a successive process^[27,28]. To be specific, first, six-coordinate Al(OH)₃ EFAL species are generated from zeolite frameworks, and then five-coordinate Al(OH)₂⁺ and four-coordinate AlOH²⁺ EFAL species are produced in sequence by eliminating one and two water molecules through interaction between Al(OH)₃ EFAL species and BAS. Strangely, no correlation between four- and five-coordinate Al species is observed in our work. These data suggest the six- and five-coordinate EFAL species are closer than the four- and five-coordinate EFAL species, which may be due to the evolution of the six-coordinate Al(OH)₃ EFAL species into the five-coordinate neutral AlO(OH) EFAL species in channels by removing one water molecule rather than the five-coordinate Al(OH)₂⁺ EFAL species interacting with BAS, that is, the four-coordinate framework Al species. After steaming at 400 °C and 500 °C, the autocorrelations of the three kinds of Al species (four-coordinate framework Al, five- and six-coordinate EFAL) and the spatial proximity between any two of them are clearly observable in the spectra [Figure 5C and D], indicating that they are in close proximity to one another. This may contribute to the complex spatial proximity of various hydroxyl groups during dealumination.

Dynamic evolution path of HZSM-5 zeolite framework under steam treatment

Based on our experiment data, the dynamic evolution path of the HZSM-5 zeolite framework under mild and severe steam conditions is proposed, as illustrated in Scheme 1, mainly including framework dealumination [Path 1-5 in Scheme 1] and defect self-healing [Path 6-8 in Scheme 1]. Under mild steam conditions (T ≤ 200 °C), almost no dealumination occurs, but a large number of hydrogen-bonded silanols (silanol nests) inevitably generated during the synthesis and calcination process are condensed to form new framework Si-O-Si bonds [Path 6 in Scheme 1]. As the steam treatment gradually becomes more severe (T ≥ 300 °C), the framework Al-O bonds undergo irreversible hydrolysis with water molecules in the channels, resulting in the dynamic evolution of the framework tetrahedral Al atoms. Certainly, partially bonded framework (SiO)_4-n-Al(OH)_n species, Al(IV)-2, generated through the partial hydrolysis of framework Al-O bonds is an important intermediate species in dealumination [Paths 1 and 2 in Scheme 1]. The newly generated bridge hydroxyl groups of Al(IV)-2 can be perturbed by the newly generated framework Al-OH groups due to hydrogen bonding interactions, which causes the ¹H chemical shifts to lower fields, appearing at 6.0-9.0 ppm and 12.0-15.0 ppm. When all four framework Al-O bonds bonded to one Al atom are completely hydrolyzed by four water molecules, the Al atom is completely removed from the framework, producing initial extra-framework Al(OH)₃·H₂O species and silicon hydroxyl nests [Path 3 in scheme 1]. The Al(OH)₃·H₂O species can subsequently migrate away from silicon nest [Path 5 in scheme 1] and transform into new EFAL species through hydroxyl condensation, such as Al(OH)₂⁺, Al(OH)²⁺, AlO⁺, Al₂O₃, dimers, and trimers, which can be absorbed near the BAS by electrostatic interactions, stabilizing the framework but blocking the channel [Path 4 in Scheme 1]. In addition, the generated silanol nest will continue to evolve into new framework Si-O-Si bonds through condensation reactions of hydroxyl groups, including the direct condensation of adjacent Si-OH groups within silanol nests [Path 6 in Scheme 1] and the migration of Si(OH)₄ species produced by desilication to the silanol nest for reorganization [Paths 7 and 8 in Scheme 1]. Notably, during dealumination of HZSM-5 zeolite frameworks, these structurally distinct Al species coexist simultaneously, which complicates the spatial proximity of the different hydroxyl groups on them. Furthermore, the site specificity for steam-induced framework Al atoms removal in HZSM-5 zeolites has attracted much attention. Many studies have shown that Al atoms located at the intersections of straight and sinusoidal channels are more susceptible to dealumination, followed by sinusoidal channels, while Al atoms facing the straight channels are most stable^{[25,26,29,34]}.

Scheme 1. Proposed dynamic evolution path of HZSM-5 zeolite frameworks under steam treatment.