1 Introduction

As the core of post-combustion removal of harmful nitrogen oxides (NOx) produced by lean-burn engines, selective catalytic reduction with NH3 (NH3-SCR) catalyst is bifunctional: Acid sites adsorb/activate NH3 molecules and redox sites promote the cycling of metal ions between high and low valence [1]. The activation of the gases at bifunctional sites occurs via two types of mechanisms: (1) the Langmuir–Hinshelwood (L–H) mechanism, where adsorbed ammonia species react with nitrites/nitrates; (2) the Eley–Rideal (E–R) mechanism, where the adsorbed ammonia species react with NOx gas [2]. Because of the difference in the adsorption/activation of NO and NH3 at the reaction sites, the low-temperature (150–400 °C) reaction mainly follows the L–H mechanism, whereas the high-temperature (400–550 °C) reaction generally proceeds via the E–R mechanism [3, 4].

Commercial NH3-SCR catalysts function under the high-temperature (T > 650 °C) hydrothermal (~ 10 vol% H2O) condition [5,6,7,8], which generally accelerates performance deterioration. Hence, high-temperature hydrothermal stability is critical. Remarkably, in previous studies [9,10,11], we observed hydrothermal deactivation to varying degrees at low (150–400 °C) and high (400–550 °C) temperatures with “seagull”-shaped NOx conversion profiles using the catalysts following hydrothermal aging. Sacrificed bifunctional sites may be related to this “seagull”-shaped profile [12,13,14]. Focusing on the hydrothermal stabilities of NH3-SCR catalysts, deep understanding of the correlation between the determined reaction sites and the deactivation mechanism at the low (150–400 °C) and high (400–550 °C) temperatures is required. However, to the best of our knowledge, there remains a lack of systematic research regarding this correlation. Thus, these sites should be altered to modify the catalytic and deactivation properties of the catalysts to investigate this relationship.

CHA zeolites (SAPO-34 and SSZ-13) are representative supports for NOx removal. The hydrothermal deactivation of SAPO-34 after extended storage at T < 70 °C is mysterious. However, Cu/SAPO-34 still displays more advantages compared with Cu/SSZ-13, such as low synthetic cost, good low-temperature activity, and excellent high-temperature hydrothermal stability. Si atoms substitute Al or P atoms in the AlPO framework, forming Si(OAl) structures. Owing to the different electronegativity of the Si and Al atoms, extra protons are required to neutralize the framework of the zeolite [15]. These protons in the Si(OAl) structures provide ion-exchange sites for anchoring copper ions and acid sites for NH3 adsorption. Thus, modifying the Si(OAl) structures may alter the reactant sites. The silicate ions depolymerized from SiO2 sols participate in the formation of Si(OAl) structures. The surfactant polyvinyl alcohol (PVA) adsorbs on the surfaces of the SiO2 sols, where decreased surface energy prevents the sols from growing into large particles, affecting the formation of the Si(OAl) structures [16,17,18]. In addition, the hydroxyl groups of PVA react with the hydroxyls located at the surfaces, pores, and channels of AlPO [17], affecting the reaction between partial silicate ions and AlPO, which also alters the Si(OAl) structures. Herein, PVA may be the second structure-directing agent affecting the Si(OAl) structures, altering the redox and acid properties of Cu/SAPO-34 [16, 19,20,21].

Combining in situ diffuse reflectance infrared Fourier transform (DRIFT) reactions and characterization reveals the specific roles of the reaction sites in hydrothermal deactivation at low (150–400 °C) and high (400–550 °C) reaction temperatures. We hope that this discovery is useful in designing an NH3-SCR catalyst with excellent hydrothermal stability over the entire reaction temperature range (150–550 °C).

2 Experimental

2.1 Preparation and hydrothermal aging of catalysts

The redox and acid properties were altered by changing the amount of PVA in the synthesis of SAPO-34. The molar composition of SAPO-34 gel is 1 Al2O3:1 P2O5:0.6 SiO2:3.5 triethylamine (TEA):50 distilled water. Upon mixing pseudo-boehmite, H3PO4, and distilled H2O, Gel A was formed. PVA (mass of 0.3, 1.0 and 1.5 g) was added to distill H2O, followed by Si sols (analytical grade, AR, Shanghai Nalco, China) and TEA. This mixture was well-stirred as Solution B. Gel A was added to Solution B, forming a new solution. This well-mixed solution was allowed to stand at 25 °C for 12 h and then poured into a stainless, Teflon-lined autoclave. Hydrothermal crystallization was then conducted for 24 h at 200 °C. The solid product was washed three times with distilled H2O and then dried completely at 105 °C. Finally, the product was treated for 6 h at 600 °C under flowing air. Copper nitrates were the precursors for preparing the Cu/SAPO-34 catalysts. All raw materials were AR reagents purchased from Chengdu Chron Chemicals (China) unless otherwise stated and used without further purification. The detailed methods are described in the section “Preparation of catalysts” in Supplementary Material.

A wash-coated catalytic monolith (Corning, USA) with approximately 120 g·L–1 was used for performance assessment. The prepared monolithic catalyst was age-accelerated at 700 °C for 10 h under flowing air containing 10 vol% H2O. These samples were denoted as SA(X) and Cat(X)-F/HT (Table 1), where SA, Cat, and X represent Cu-free SAPO-34 zeolite, Cu/SAPO-34 catalyst, and the mass of PVA, respectively; F and HT represent fresh and hydrothermally aged catalysts, respectively. After hydrothermal aging, the catalysts exhibited hydrothermal deactivation, except Cat(0) without PVA. The performance deterioration at 150–400 and 400–550 °C is described as hydrothermal deactivation in the low- and high-reaction regions, respectively.

Table 1 Amounts of PVA and descriptions of samples
Full size table

2.2 Physical and chemical characterization

X-ray photoelectron spectroscopy (XPS) was used to detect the surface concentrations and inductively coupled plasma-atomic emission spectrometry (ICP-AES, IRIS Advantage, USA) revealed the Cu content, as shown in Table 2. The in situ DRIFT spectroscopy using a Nicolet 6700 (Thermo Fisher Scientific, USA) was conducted to study the possible mechanisms in the deactivation regions. The temperature-programmed reduction of hydrogen (H2-TPR) and XPS was conducted to characterize the copper species, and the temperature-programmed desorption of NH3 (NH3-TPD) was performed to determine the acid sites. The structural properties were investigated using N2 physisorption, X-ray diffraction (XRD), scanning electron microscopy (SEM), DRIFT spectroscopy, and 27Al solid-state nuclear magnetic resonance (NMR) spectroscopy. The detailed experimental procedures and parameters are listed in the section “Characterization” in Supplementary Material.

Table 2 Molar ratios of Si/(Al + P) and Cu contents
Full size table

2.3 Catalytic activity

The as-prepared monolithic catalysts were evaluated in NH3-SCR from 150 to 550 °C. Before the study, pretreatment was performed under the mixed reactant gas balanced using N2 with a total gaseous hourly space velocity of 40,000 h–1. The pretreatment time and temperature were 1 h and 550 °C, respectively, and the mixed gas contained 220 × 10–6 NH3, 200 × 10–6 NO, 10 vol% O2, and 5 vol% H2O. Using the gas concentrations monitored with gas Fourier transform infrared spectroscopy (FTIR, Nicolet Antaris IGS-6700, Thermo Fisher Scientific), the NOx conversion is obtained at intervals of 50 °C using the following equation:

$${\text{NO}}\;{\text{ conversion}} = \left( {1 - \frac{{\left[ {{\text{NO}}} \right]_{{{\text{out}}}} + \left[ {{\text{NO}}_{2} } \right]_{{{\text{out}}}} + 2\left[ {{\text{N}}_{2} {\text{O}}} \right]_{{{\text{out}}}} }}{{\left[ {{\text{NO}}_{x} } \right]_{{{\text{in}}}} }}} \right) \times 100 \%$$
(1)

where [NO]out, [NO2]out, and [N2O]out are the outlet concentrations (× 10–6) of NO, NO2, and N2O, respectively; [NOx]in represents the total concentration (10–6) of nitrogen oxides within the reactor.

3 Results and discussion

3.1 Properties of PVA-modified and unmodified samples

XRD patterns and morphologies of all Cu-free zeolites (Fig. 1a) reveal diffraction peaks and cubic phases that correspond to those of SAPO-34. Figure 1b and inset show the 27Al solid-state NMR spectra and the integrated signal areas of the signals representing the framework tetra-coordinated Al (AlIV) atoms, respectively. The water-soluble PVA at a low concentration (0.3 g) uniformly disperses into the zeolite precursors in the form of random coils, which are independent of each other because of extremely weak interactions [22]. The PVA coils enter the pores and channels, adsorbing on the surface of AlPO, assisting in the formation of pores and channels on SA(0.3), which yields a large Brunauer–Emmett–Teller surface area (SBET) of 856 m2·g–1 (Table 3). In addition, PVA prevents the aggregation of the SiO2 sols via hydrogen bonding. Overall, 0.3 g PVA improves the crystallinity of SAPO-34 and increases the content of framework AlIV atoms by 11.6% [23]. The interaction strengthens with PVA concentration increasing, inducing the formation of three-dimensional (3D) networks that may fix precursors [21, 22, 24]. For an intermediate concentration of PVA (1.0 g), the 3D network and coils are observed, decreasing the content of AlIV by 7.3%. In addition, the network fixes the crystal nucleus akin to the colloidal stabilizer, inhibiting the attachment of the precursors to the crystals [25, 26]. Thus, small and uniform crystals with sizes of 2–3 μm are formed on SA(1.0), which are smaller than the crystals of 6–7 μm on SA(0).

Fig. 1
figure 1

a XRD patterns and (insets) SEM images of zeolites; b 27Al solid-state NMR spectra and signal areas of AlIV of Cat(0)-F, Cat(0.3)-F, Cat(1.0)-F, and Cat(1.5)-F; c NH3-TPD profiles of zeolites; d quantified areas of NH3-TPD

Full size image
Table 3 Texture properties of zeolite
Full size table

When the concentration of PVA is increased to 1.5 g, PVA mainly forms a large network. SAPO-34 exhibits a microporous structure, with pore sizes of generally < 2 nm. Thus, it is challenging for the large PVA network to enter the channels and pores of the SAPO-34 zeolite, resulting in less influence on the internal channels and pores. This is consistent with the similar textural properties of SA(0) and SA(1.5) (Table 3). Moreover, SA(1.5) consists of 4–6 μm crystals, which are also similar to those of 6–7 μm on SA(0). However, the barrier of the PVA network prevents the reaction between the silicate ions and AlPO, decreasing the content of AlIV by 9.1%. The NH3 desorption and quantified areas shown in Fig. 1c, d demonstrate that the unmodified SA(0) exhibits the largest moderate- (MP) and high-temperature peaks (HP) at 260 and 345 °C, which are ascribed to moderate and strong acid sites [27], respectively. The SA(1.0) zeolite shows the largest low-temperature peak (LP) at 186 °C, which is assigned to weak acid sites [28].

Therefore, different amounts of PVA successfully alter the properties of the SAPO-34 zeolites, which should further affect the reaction sites on the subsequently prepared Cu/SAPO-34 catalysts, with various hydrothermal durabilities at a high temperature (700 °C).

3.2 NH3-SCR performance

Cu/SAPO-34 catalysts were prepared by the incipient wetness impregnation method using the prepared SAPO-34. The catalytic performance was then evaluated in triplicate, and the data with confidence intervals are shown in Fig. 2. Four types of hydrothermal deactivation in the reaction regions are successfully detected. The NH3-SCR performance is retained on Cat(0)-HT. The NOx conversion on Cat(0.3)-HT declines by > 7% in the low-reaction (150–400 °C) region and by > 10% in the high-reaction (400–550 °C) region. The hydrothermal deactivation of Cat(1.0)-HT is observed in the high-reaction region, with NOx conversion decreasing by > 10%. The hydrothermal deactivation of Cat(1.5)-HT is observed in the low-reaction region, with the maximum loss of NOx conversion reaching 6%. The amount of N2O obtained as the by-product (Fig. S1) is < 7 × 10–6, with no changes even after hydrothermal aging.

Fig. 2
figure 2

NH3-SCR performances of a Cat(0)-F/HT, b Cat(0.3)-F/HT, c Cat(1.0)-F/HT, and d Cat(1.5)-F/HT

Full size image

3.3 Possible reaction mechanisms in hydrothermal deactivation region

The surface molar ratios of Si/(Al + P) on fresh and hydrothermally aged catalysts are 0.03, with Cu contents also unchanged after hydrothermal aging (Table 2). Therefore, the hydrothermal deactivation is unrelated to the elemental compositions. In situ DRIFT reaction analysis of the pre-adsorbed species and gaseous reactants was conducted to investigate the possible reaction pathways in the hydrothermal deactivation region. NOx conversion on Cat(0.3)-HT declines by ~ 7% at 200 °C and 11% at 450 °C, and therefore, Cat(0.3)-F/HT was used as representative examples. Furthermore, 200 and 450 °C were used as the evaluation temperatures. The in situ DRIFT reactions on Cat(0.3)-F/HT are shown in Fig. 3, and the results of the other catalysts are shown in Figs. S2–S4. The top panel of Fig. 3 shows that after Cat(0.3)-F/HT saturation with NH3 gas at 200 °C, bands attributed to NH4+ species adsorbed at Brønsted acid sites (BAS) are observed at ~ 1463, 1717, 3278, and 3369 cm–1 [29,30,31]. Further, bands at ~ 1614 and 3187 cm–1 are assigned to NH3 species coordinated at Lewis acid sites (LAS) [32,33,34]. The bands decrease while introducing NO + O2 into the saturated catalyst by NH3, suggesting the participation of the absorbed ammonia species in the catalytic reaction. The band at ~ 1614 cm–1 related to Cu2+ is attributed to NH3 species initially and then nitrate at ~ 1618 cm–1 as the reaction proceeds [35].

Fig. 3
figure 3

In situ DRIFT reactions between pre-adsorbed species and gaseous reactants on a Cat(0.3)-F (200 °C), b Cat(0.3)-HT (200 °C), c Cat(0.3)-F (450 °C) and d Cat(0.3)-HT (450 °C) at 1, 3, 5 or 10 min (top part of every panel representing spectra exposed to NH3 followed by NO + O2 for various time; bottom part representing spectra exposed to NO + O2 followed by NH3 for various time)

Full size image

While introducing the mixture of NO and O2 at 200 °C, only a band at 1622 cm–1 is attributed to the bridged nitrate generated at the substituted Cu sites [36, 37]. After nitrate consumption by NH3, adsorbed ammonia species accumulate again on the surface. Adsorbed ammonia species and bridged nitrate cooperate in the catalytic reaction at 200 °C, corresponding to the L–H mechanism. Additionally, no adsorbed nitrate at 450 °C induces the E–R mechanism, where adsorbed ammonia species are the critical active intermediates. After hydrothermal aging, the same adsorbed species are observed over Cat(0.3)-F/HT, demonstrating the unchanged mechanism in the hydrothermal deactivation regions. The trend observed over the other catalysts also verifies this view (Figs. S2–S4). Besides, the rate and apparent activation energy (Ea) were calculated according to the formulas in Supplementary Material. Ea before and after hydrothermal aging is calculated based on the kinetic data, as shown in Fig. S5. Because Ea is related to active species, the similar values confirm the unchanged mechanism after hydrothermal aging [38].

3.4 Hydrothermal deactivation in low-reaction (150–400 °C) region

3.4.1 Redox sites

The performance deteriorations on Cat(0.3)-HT and Cat(1.5)-HT are centered in the low-reaction (150–400 °C) region. The isolated Cu2+ and surface oxygen species participate in the redox cycle of the NH3-SCR reaction. Figure 4a depicts the results of H2-TPR. Cat(0.3)-HT and Cat(1.5)-HT exhibit decreases in isolated Cu2+ by 7.1 and 13.6 μmol·g–1, respectively. This is consistent with the changed Region I observed in the ultraviolet–visible spectra (Fig. S6a).

Fig. 4
figure 4

a H2-TPR (gray and red curves representing F and HT catalysts, respectively) and change in isolated Cu2+ and b O 1s XPS profiles of Cat(0)-F/HT, Cat(0.3)-F/HT, Cat(1.0)-F/HT and Cat(1.5)-F/HT

Full size image

Figure 4b shows O 1s regions in the XPS profiles, with peaks attributed to surface chemisorbed oxygen Oβ (denoted as Oβ1 and Oβ2 at 533.1 and 534.9 eV, respectively), and lattice oxygen O2– (denoted as Oα at 531.7 eV) [30, 39]. The Oβ1 species include O–, O22– and O2– [40, 41], and Oβ2 represents the oxygen within the hydroxyl groups of Si–OH–Al or oxide defects [42, 43]. Oβ species easily transfer electrons between the surface adsorbed dioxygen and the gaseous oxygen, exhibiting higher mobility than Oα species [44]. Herein, Oβ species are generally the critical oxygen species for oxidizing NO, greatly contributing to the catalytic efficiency via the “fast NH3-SCR” reaction [42]. As listed in Table 4, the percentage of Oβ in total O species (Oβ/Otol) decrease from 0.53 on Cat(0.3)-F to 0.49 on Cat(0.3)-HT and 0.51 on Cat(1.5)-F to 0.49 on Cat(1.5)-HT. In addition, the surface chemisorbed oxygen species weaken the negative influence of the 0.41 wt% decrease in the surface isolated Cu2+ from Cat(1.0)-F to Cat(1.0)-HT (Table 4 and Fig. S6a) [44, 45].

Table 4 Concentrations of surface species and textural properties
Full size table

Thus, isolated Cu2+ and Oβ species collectively determine the hydrothermal deactivation in the low-reaction (150–400 °C) region. The bridged nitrates are involved in the catalytic reaction in this region based on the L–H mechanism, with the NOx emission control (deNOx) easily affected by the redox sites (the isolated Cu2+ and surface chemisorbed oxygen).

3.4.2 Acid sites

The alkaline NH3 molecules probe the acid sites. The NH3 desorption profiles, as shown in Fig. 5a, are well-fitted into three regions (LP, MP, and HP). As demonstrated by the integrated areas (Fig. 5b), LP on Cat(0.3)-HT decreases by ~ 23.2%, showing a severe performance deterioration in the low-reaction (150–400 °C) region. Furthermore, Cat(1.5)-HT also exhibits a severe deactivation in this region, with an LP loss of 5.7%. Generally, LP is attributed to weak acid sites [46], and thus, the loss of weak acid sites accelerates the performance deterioration in this region.

Fig. 5
figure 5

a NH3-TPD and b decline in acid sites after hydrothermal aging of Cat(0)-F/HT, Cat(0.3)-F/HT, Cat(1.0)-F/HT and Cat(1.5)-F/HT

Full size image

3.5 Hydrothermal deactivation in high-reaction (400–550 °C) region

The performances of Cat(0.3)-HT and Cat(1.0)-HT deteriorate in the high-reaction (400–550 °C) region. However, Cat(1.0)-HT still bears the newly generated redox sites (Fig. 4). In addition, the NH3 oxidation conversion and product distribution (upper-left and lower-right insets in Fig. S7, respectively) suggest no aggravated NH3 oxidation after hydrothermal aging. Hence, the redox sites (copper species and Oβ) are independent of the performance deterioration in this region [47,48,49].

3.5.1 Acid sites

The declines in MP are 13.4% and 28.1% on Cat(0.3)-HT and Cat(1.0)-HT, respectively. Moreover, both catalysts exhibit considerable decreases in HP by 8.4% and 6.7%, respectively. Generally, MP and HP are regarded as strong LAS and strong BAS associated with the zeolite framework, respectively [49, 50]. The in situ DRIFT reactions, as shown in Fig. 3, also confirm the presence of LAS and BAS in the high-reaction (400–550 °C) region. Therefore, both acid sites determine the hydrothermal deactivation in this region.

3.5.2 Relationship between structural stability and acid sites

The decreased diffraction peak (Fig. S8) is generally related to the collapse of the structure [51,52,53]. According to the peak intensity at 2θ = 9.5°, Cat(0)-HT and Cat(1.5)-HT show 5% and 10% loss, respectively, which are much lesser than that observed over Cat(0.3)-HT (20%) and Cat(1.0)-HT (19%), as shown in Fig. 6a. In addition, the textural properties (Table 4) show that SBET decreases by 9.6% and 6.2% on Cat(0)-HT and Cat(1.5)-HT, respectively, which are also much lower than those on Cat(0.3)-HT (43.7%) and Cat(1.0)-HT (19.1%). The diffraction intensity and SBET reveal the poor structural stabilities of Cat(0.3)-HT and Cat(1.0)-HT, which is consistent with the changes in their morphologies, as shown in Fig. S9. The cleavage of the framework Si(OAl) structures results in the structural collapse of Cu/SAPO-34. As shown in Fig. 6b, bands at 3628 and 3561 cm–1 represent the stretching vibrations of Si(OAl) structures located in the eight-member and double six-member rings, respectively [54,55,56,57]. Few changes in the intensities in the spectra of Cat(0)-HT and Cat(1.5)-HT suggest that most Si(OAl) remains intact, consistent with their stable structures. The signals representing the Si(OAl) structures clearly deteriorate in the DRIFT spectra of Cat(0.3)-HT and Cat(1.0)-HT, also demonstrating their poor structural stabilities. 27Al solid-state NMR spectroscopy (Fig. S10) is a powerful tool for analyzing the framework AlIV atoms. The loss results (Fig. 6c) show that AlIV species decrease by 4.7%, 19.7%, 38.7%, and 34.1% on Cat(0)-HT, Cat(1.5)-HT, Cat(0.3)-HT, and Cat(1.0)-HT, respectively. The two former samples lose lower concentrations of AlIV, with the Si(OAl) structures undergoing attenuated hydrolysis during hydrothermal aging. The weaker hydrolysis is consistent with the superior structural integrities of Cat(0)-HT and Cat(1.5)-HT compared to those of Cat(0.3)-HT and Cat(1.0)-HT.

Fig. 6
figure 6

a Intensities and losses of main diffraction peaks, b DRIFT spectra of OH vibrational regions after NH3 adsorption at 50 °C, and c lost AlIV species after hydrothermal aging based on 27Al solid-state NMR spectroscopy of Cat(0)-F/HT, Cat(0.3)-F/HT, Cat(1.0)-F/HT and Cat(1.5)-F/HT

Full size image

The framework Si(OAl) structures deteriorate to extra-framework Alβ species [58]. As demonstrated by the results listed in Table 4 based on the Al 2p XPS profiles (Fig. S11), the Alβ species on Cat(0.3)-HT and Cat(1.0)-HT increase by 1.1 wt% and 2.3 wt%, respectively, which are clearly higher than those (0.7 wt% and –3.1 wt%) on Cat(0)-HT and Cat(1.5)-HT, respectively. Thus, the structural stability is closely related to the strong LAS/BAS, accelerating the performance deterioration in the high-reaction (400–550 °C) region.

4 Conclusion

This work investigated the critical roles of various reaction sites on hydrothermal deactivation in the low- (150–400 °C) and high-reaction (400–550 °C) regions over Cu/SAPO-34. First, the hydrothermal deactivation in these two regions was successfully manipulated over Cu/SAPO-34 using PVA. Then, the deactivation region was analyzed in terms of the detailed roles of the redox/acid sites. The results verified that the catalytic reactions at the low (150–400 °C) and high (400–550 °C) reaction temperatures still predominantly follow the L–H and E–R mechanisms, respectively, even after hydrothermal deactivation. However, the key difference on the reactant sites results in “seagull”-shaped NOx conversion profiles over Cu/SAPO-34 after hydrothermal aging, with an inflection point at 400 °C. The low-reaction region requires adsorbed nitrates/ammonia species, and therefore, hydrothermal deactivation in this region can be suppressed by optimizing the easily reducible isolated Cu2+, surface chemical oxygen, and weak acid sites. The adsorbed/activated ammonia species dominate in the high-reaction region, with restrained performance deterioration due to the increased stability of the moderate/strong acid sites related to the structure. This work provides several theoretical insights to optimize the hydrothermal stabilities of the Cu/zeolite catalysts.