S-compounds in transportation fuels can cause acid rain and release PM 2.5. Hence, it is increasingly important to perform ultra-deep desulfurization, to alleviate environment pollution. The Environmental Protection Agency has proposed a strict regulation, in which the sulfur content should be reduced to 10 ppm [1, 2]. However, by conventional hydrodesulfurization (HDS), it is difficult to remove the sulfur from fuels and satisfy the new regulation for gasoline without loss of the octane number [3]. FCC gasoline contains about 20–40 wt% olefins [4], which may provide a fairly good octane number. Olefins can easily be reacted with hydrogen in the hydrodesulfurization reaction. In order to maintain a high octane number, the technology and catalyst should exhibit good selectivity during desulfurization. Researchers are trying to achieve maximum sulfur removal with minimum olefin saturation. Therefore, a novel ultra-deep desulfurization technology for gasoline is important to prevent the loss of octane number when upgrading FCC gasoline. Over the years, researchers have developed new methods such as adsorption [5, 6], oxidation [7], extraction [8], and biodesulfurization [9], for the ultra-deep desulfurization of products. The advantages of adsorption and reactive adsorption, such as high desulfurization, high selectivity, and low energy consumption, can aid in developing the technology and catalyst. A reactive adsorption desulfurization (RADS) technology, called the "S-Zorb" technology, was developed and industrialized by Conoco Phillips Petroleum Co, and Ni/ZnO adsorbents were used in the fluidized bed to produce low-sulfur gasoline during the S-Zorb process [10, 11]. However, frequent regeneration decreases the sulfur saturation adsorption capacity of ZnO. More importantly, the Ni-based catalyst was always used at a high space velocity to control the selectivity of the RADS, and could not reach the level required for ultra-deep desulfurization. Therefore, it is important to develop a novel RADS catalyst and technology for inhibiting olefin saturation. There have been some efforts over the years to design and develop the desulfurization technology and catalyst to avoid the loss of octane number. In our previous research [12], Cu/ZnO adsorbent exhibited good desulfurization activity, stability, and selective hydrogenation in a fixed-bed reactor for RADS. Cu- based catalysts are used in many fields, such as methanol steam reforming [13], water-gas shift reaction [14], and so on. Active metal Cu has a low melting point and Tammann temperature, and is prone to migration to the surface of metallic particles [15], which leads to good selectivity. Wang et al. [16] investigated the adsorption of thiophene on active metal (100) surfaces by using the density functional theory (DFT), and suggested that Cu-based materials could be the next generation of adsorbents for RADS.

Nowadays, Cu-based adsorbents face issues with the sulfur saturation adsorption of ZnO and stability in a fixed-bed reactor. Metallic Cu particles may be easily sintered in the reaction process; however, ZnO and Al₂O₃ can be regarded as stabilizers to prevent the Cu particles from sintering in the Cu crystallites [17, 18], to thereby improve the activity and stability. The dynamic capacity of ZnO-based adsorbents and the reactive activity are always dependent on the particle size [19, 20] of ZnO and Cu, respectively. ZnO not only acts as a support, but also as an S-acceptor [21]. ZnO has been used widely in the catalytic process, and the properties of ZnO, such as particle size, quanfinement [22, 23], and morphology [24] influence the catalytic performance [25]. Therefore, many nanostructured ZnO have been synthesized [26, 27]. An active component, when supported on a porous carrier such as alumina, can stabilize small particles by the commonly used methods [28]. The pore volume can increase in dynamic capacity to promote the reaction of H₂S with ZnO, at the depth of the alumina matrix. Alumina is widely used as a catalyst or catalyst support binder for a variety of applications [29]. Ordered mesoporous alumina has higher surface areas, larger pore sizes, ordered pores, and better stability than bulk alumina materials [30]. The high surface area favors the dispersion of active metals and enhances molecular interdiffusion. Tian et al. [31] reported that the matrix of ordered mesoporous alumina could improve the dispersion of Ni nanoparticles, which is promising for long-term stability tests.

To enhance the sulfur adsorption of Cu-based adsorbents for reactive adsorption desulfurization, ordered-mesoporous- structure Cu-ZnO-Al₂O₃ adsorbents were designed and synthesized by a one-pot evaporation-induced self-assembly (EISA) strategy. The ordered mesoporous structures provide a high thermal stability, and large specific surface areas and pore volumes. Therefore, the ordered-mesoporous-structure Cu- ZnO-Al₂O₃ adsorbents are beneficial for transforming S-compounds to ZnO; moreover, they exhibit larger sulfur saturation capacity than unordered mesoporous structures.

Zinc acetate dehydrate (Zn(AC)₂·2H₂O), copper nitrate trihydrate (Cu(NO₃)₂·3H₂O), aluminum isopropoxide (Al(OiPr)₃), nitric acid (HNO₃), and ethyl alcohol (C₂H₅OH) were provided by Sinopharm Chemical Reagent Company (China). The block copolymer poly (ethylene glycol)-block-poly (propylene glycol)-block-poly-(ethylene glycol) (Pluronic P123, PEG-20-PPG- 70-PEG-20) was purchased from Sigma-Aldrich (Germany). All other reagents (Grade AR) were purchased and used without any further purification.

In a one-pot EISA synthesis process [32], P123 (2.1 g) was dissolved in ethyl alcohol, followed by the addition of 67 wt% HNO₃ (2.3 mL) and 4.08 g aluminum isopropoxide Al(OPri)₃. Zinc acetate dehydrate Zn(AC)₂·2H₂O and copper nitrate trihydrate Cu(NO₃)₂·3H₂O were stirred vigorously at atmospheric temperature, and left for more than 24 h. The obtained mixture was then placed in an oven at 60 ℃ to evaporate ethanol. The precursors were calcined at 500 ℃ for 4 h at a heating rate of 1 ℃/min. The ZnO content was varied as 0, 5, 15, 20, and 25 wt%. The CuO content was fixed at 10 wt% in all adsorbents in this work. The series of adsorbents are denoted as Cu-xZnO- Al₂O₃, where x represents the percentage ZnO content.

The samples were characterized by X-ray powder diffraction (XRD) to obtain information on the structural properties of the adsorbents, using a PANalytical instrument with Cu K_α radiation at a scan step of 0.02° and 2°.

N₂ adsorption-desorption experiments were performed on a ChemBET 2000 instrument (Quantachrome, USA). Nitrogen was used as the adsorption agent at –196 ℃. Brunauer-Emmett-Teller (BET) surface areas were calculated from the isotherms, and the total pore volumes were calculated by the Barrett-Joyner-Halenda (BJH) method. The BJH pore sizes were calculated using the desorption branch of the isotherms.

Temperature-programmed reduction by H₂ (TPR) was carried out to assess the reducibility of the Cu-ZnO-Al₂O₃ adsorbents. Approximately 100 mg of each catalyst loaded into a U-shaped quartz reactor and heated up to 650 ℃ in a flow of 10% H₂/Ar gas mixture at a heating rate of 10 ℃/min. The consumption of H₂ was monitored with a TCD. The average volume-surface diameter and dispersion of Cu were calculated according to previously reported equations [33, 34]. The dispersion and area of surface Cu were calculated according to the following equations:

(1)

(2)

In these equations, N_av is Avogadro's constant, M_cu is the relative atomic mass of copper (63.46 g/mol), and 1.4×10¹⁹ is the number of copper atoms per square meter, because the average surface area of a copper atom is 7.11×10^–2 nm².

The average volume-surface diameter can be expressed as:

(3)

where ρ_cu is the density of copper (8.92 g/cm³).

Transmission electron microscopy (TEM) images of the samples were acquired using a JEOL JEM-2100 UHR microscope. The sample particles were first ground and then suspended in alcohol in an ultrasonic bath, following which they were placed in a Cu cellulose-coated grille.

Thermogravimetry (TG) was used to investigate the thermal behaviors of the Cu-ZnO-Al₂O₃ precursors. The TG experiments were carried out in air at a heating rate of 20 ℃/min.

The composition of the hydrocarbon groups in gasoline was determined with an Agilent 7890N gas chromatograph with a PONA capillary column. Quantitative analysis of the total sulfur concentration in model gasoline was performed by a Multi EA 3100 S/N trace analyzer. The detection range was 0.1–1000 mg/L.

The desulfurization performances of the adsorbents for model gasoline upgradation were evaluated in a fixed-bed reactor. The oxidized states of the adsorbents (approximately 1 g) were reduced in the constant-temperature zone of the reactor, which was embedded between the quartz sand beds. First, in situ activation was performed in hydrogen flow at 1 MPa for 6 h. After cooling to 400 ℃ at 1 MPa, the model gasoline was fed into the reactor with a syringe pump at a predetermined flow rate, a H₂/feed volumetric rate of 200/1, and LHSV of 1 h^–1.

The activity of the adsorbent was estimated using RADS (amount of reacted thiophene):

(4)

where C_S⁰ is the thiophene content in the feedstock (wt%) and C_s is the thiophene content in the products (wt%).

(5)

The breakthrough sulfur capacity was determined as follows:

(6)

where q_breakthrough is the breakthrough sulfur capacity of the adsorbent (mg/g), v is the feed volumetric flow rate (mL/min) at any time t (min), and m is the weight of the adsorbents (g). The breakthrough time was defined as the RADS time when the sulfur concentration of the effluent-desulfurized feed exceeded 10 mg/L.

The ZnO conversion to ZnS (X) at the sulfur breakthrough point (10 mg/L) is expressed as [20]:

(7)

N₂ adsorption-desorption isotherms of the samples synthesized by the one-pot EISA method are shown in Fig. 1. The Al₂O₃ sample shows typical type-Ⅳ curves with the H1 hysteresis loop, attributed to ordered cylindrical mesoporous channels [35]. The hysteresis loops at a higher relative pressure (p/p₀) from 0.5 to 0.9 indicate the uniform mesoporosity of the Al₂O₃ sample. The shapes of the isotherms and loop hysteresis with the proper addition of Cu and Zn species are similar to those of the Al₂O₃ sample, whereas the capillary condensation of N₂ of the adsorbents are diminished. These results indicate that the ordered mesoporous structure was not destroyed by the addition of an appropriate amount of Cu and Zn species. However, for a ZnO content of 25 wt%, the isotherm exhibited typical type-Ⅳ curves with the H4 hysteresis loop. This indicates that the ordered mesoporous structure was blocked and destroyed with the addition of ZnO particles. The pore distribution of the Cu-ZnO-Al₂O₃ adsorbents is presented in Fig. 1. (2), and their detailed characteristics are presented in Table 1. The results suggest that the one-pot EISA method can form Cu-ZnO-Al₂O₃ with ordered mesoporous structures, high surface areas, and large pores.

The small-angle XRD (SXRD) and WXRD patterns of the samples are shown in Fig. 2. The Al₂O₃ sample displays two peaks at 2θ = 1.27° and 2.18°, which are attributed to the (100) and (110) planes, respectively. The results indicate that the hexagonally ordered mesoporous structure Al₂O₃ with a space group of p6mm symmetry was successfully synthesized [32, 35, 36]. The samples with Zn and Cu species also exhibit obvious (100) diffraction, suggesting that the samples retain their ordered mesoporous structures, which implies that these structures were not destroyed by the impregnation of appropriate amounts of Zn and Cu. However, when 25 wt% Zn was added, the obvious (100) diffraction disappeared, implying that the ordered mesoporous structure was destroyed. Thus, ordered mesoporous CuO-ZnO-Al₂O₃ adsorbents were successfully synthesized by the one-pot EISA method; this agrees with the N₂ adsorption-desorption results. To further investigate the structure of the samples, WXRD patterns for the calcined and reduced samples are shown in Fig. 2(b) and Fig. S1, respectively. All the samples were calcined at 500 ℃ and reduced at 400 ℃. The WXRD pattern of Al₂O₃ exhibits no obvious peaks, suggesting that the sample has amorphous walls. Besides, no peaks of the ZnO and CuO phases with alumina are detected in Fig. 2(b). This implies that the ZnO and CuO particles are small in size or are well dispersed. However, for Cu-25ZnO-Al₂O₃, there are some diffraction peaks. The diffraction peaks due to the ZnAl₂O₄ phase (JCPDS Card 01-071-09681) appear at 2θ = 31.3°, 36.8°, 55.6°, 59.3°, and 65.2°. The peaks at 2θ = 31.4°, 37.0°, 45.1°, 59.6°, and 65.7° are attributed to CuAl₂O₄ (JCPDS Card 00-001-1153). During calcination, strong metal-support interaction (SMSI) is favored, to form the CuAl₂O₄ and ZnAl₂O₄ phases, which destroys the ordered mesoporous structures; this agrees with the results of N₂ adsorption-desorption. There are no obvious peaks in Fig. S1(1). In Fig. S1(2)–(6), the peaks at 2θ = 43.3° and 50.4° are due to the (111) and (200) planes of Cu (JCPDS Card 00-085-1326). The WXRD patterns of the reduced samples did not exhibit the ZnO phase, indicating that the ZnO species were incorporated into the Al₂O₃ matrix. There are no diffraction peaks for the Cu phase in Fig. S1(7), indicating that CuAl₂O₄ could not be reduced to the Cu phase at 400 ℃.

TEM images of the ordered mesoporous samples are shown in Fig. 3. The typical ordered mesoporous Al₂O₃ is obvious, with the regular alignment of cylindrical pores (Fig. 3(a)), which is in accordance with the results of N₂ adsorption and SXRD. Furthermore, the ordered mesoporous structure is maintained even with an increase in the Cu and Zn contents, as shown in Fig. 3(b) and (c). However, the mesoporous structure was blocked by 25 wt% ZnO (Fig. 3(d)). These results may fit with the characteristics observed in the N₂ adsorption-desorption and XRD. Fig. 3(e) shows a fine and good dispersion of copper particles on the surface of the supports. The reduction process did not destroy the ordered mesoporous structure. In the HRTEM image, the reduced Cu-20ZnO-Al₂O₃ catalyst shows a spacing of 0.21 nm, corresponding to the Cu (111) plane in Fig. 3(f). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDS) elemental mapping of the ordered mesoporous Cu-20ZnO-Al₂O₃ adsorbent reveals that Cu, Zn, Al, and O elements are homogeneously distributed over the adsorbent (Fig. 3).

TG curves (Fig. 4) show the weight loss of the as-made samples in air atmosphere, which were obtained to investigate the decomposition behavior. The weight loss at around 110–125 ℃ is attributed to the desorption of physically adsorbed water and organic molecules [37]. Two significant weight loss steps at around 130 ℃ are due to the decomposition of P123. The third weight loss step at approximately 280 ℃ is due to the decomposition of carbonaceous and nitrogenous species. The pronounced weight loss at around 336 ℃ is attributed to the dihydroxylation during transformation into alumina [38]. There was no pronounced weight loss above 400 ℃. These results demonstrate that P123 and other groups are readily removed by calcination at 500 ℃. Moreover, the ordered mesoporous Cu-ZnO-Al₂O₃ adsorbents exhibit no weight-loss peaks for a temperature as high as 550 ℃, indicating their high thermal stability.

The differences in the TPR profiles could be interpreted as the change in the ZnO contents, for which the results are shown in Fig. 5. According to previous H₂-TPR studies of copper oxide, the reduction peak is generally at around 250 ℃ [39]. The reduction peaks of highly dispersed CuO species interacting with the support occur at a low temperature of approximately 330 ℃ for low ZnO content. The reduction peaks of CuO species with a stronger interaction with the support are due to the second peak at 355 ℃. However, broad reduction profiles at around 250–550 ℃ were observed in the Cu-25ZnO-Al₂O₃ catalyst. Gaussian fitting was used for further investigation of the Cu species. The low-temperature reduction at 345 ℃ is attributed to the reduction of CuO species. The high-temperature reduction at 472 ℃ is attributed to the reduction of the CuAl₂O₄ phase [40]. Notably, the CuAl₂O₄ and ZnAl₂O₄ phases appear in the unordered mesoporous sample at the same calcination temperature, indicating that it is more easy to form CuAl₂O₄ and ZnAl₂O₄ in the unordered mesoporous sample than in the ordered mesoporous sample [41], because of the weak support-metal interactions of the ordered mesoporous samples, due to fewer –OH groups and vacancies over the surface of amorphous Al₂O₃ [42].

The results of copper dispersion and the average diameter during N₂O oxidation and the following H₂ titration are shown in Table 1. The dispersion of copper during the reduction of Cu-ZnO-Al₂O₃ adsorbents decreased from 44% to 15% with increasing ZnO content. The average diameter of copper particles in the ordered mesoporous structure was 2.3, 2.7, 3.8, 4.2, and 4.8 nm, with increasing ZnO content; however, that of the unordered-mesoporous-structure Cu-25ZnO-Al₂O₃ adsorbent was 6.7 nm. The results show that the ordered mesoporous sample possessed larger pores to disperse the copper particles, as compared to the unordered mesoporous sample. These results agreed with the N₂ adsorption-desorption, XRD, and TEM results.

Combining all results for the characterization of the adsorbent, the possible formation mechanism of the ordered mesoporous Cu-ZnO-Al₂O₃ catalyst is illustrated in Fig. 6. The EISA method is assigned to the liquid-crystal template pathway [43] for non-siliceous mesoporous materials [44, 45]. In our synthesis system, the precursor molecules of alumina, zinc oxide precursor, and copper oxide around the P123 micellized, and self-assembly simultaneously occurred, during evaporation from an ethanol solution at 60 ℃, resulting in the formation of the p6mm hexagonal symmetry mesoporous structure. The ordered mesoporous structure with concentric mesopores was caused by the directing agent P123 [46]. The stable ordered mesoporous was successfully synthesized after calcination in air to remove the templates, following which an ordered mesoporous skeleton was formed, with the alumina, zinc oxide, and copper oxide nanoparticles dispersed well across the surface of the mesoporous skeleton.

Before evaluating the desulfurization performance of the ordered-mesoporous-structure Cu-ZnO-Al₂O₃ adsorbents, that of the traditional Cu-20ZnO-Al₂O₃-C adsorbent was evaluated for FCC gasoline in a fixed-bed reactor, to compare the characteristics. The properties of the Cu-ZnO-Al₂O₃-C adsorbent are shown in Fig. S2. The properties of FCC gasoline and products are listed in Table S1. Both, the Cu-20ZnO-Al₂O₃-C and Cu-20ZnO-Al₂O₃ adsorbent, have less olefin saturation to avoid the loss of the octane number.

In order to investigate the role of the ordered mesoporous structure in enhancing the sulfur capacity, the experiments for desulfurization performance were performed in a fixed-bed micro reactor, and the results are shown in Fig. 7. The Cu-ZnO-Al₂O₃ adsorbents had very high activities, and the sulfur content could be reduced from 100 to below 10 μg/g, thus achieving the level required for ultra-deep desulfurization. The ordered mesoporous Cu-ZnO-Al₂O₃ adsorbents exhibited better desulfurization activities than the unordered mesoporous structure. The ordered-mesoporous-structure Cu-20ZnO-Al₂O₃ adsorbents exhibited a large sulfur capacity (49.4 mg/g) with increasing ZnO content. However, the sulfur capacity of the Cu-25ZnO-Al₂O₃ adsorbent sharply decreased at 36.1 mg/g. The adsorbent Cu-20ZnO-Al₂O₃-C prepared with the common method presents a low sulfur capacity of 13.5 mg/g. These results imply that the ordered mesoporous Cu-ZnO-Al₂O₃ adsorbents have larger sulfur capacity. A series of characterization results showed that ZnAl₂O₄ existed in the Cu-25ZnO-Al₂O₃ catalyst, indicating that there is an inactive component in the RADS process. The ordered mesoporous skeleton of ZnO and Al₂O₃ was destroyed by ZnAl₂O₄, leading to the low dispersion of ZnO and preventing sulfur diffusion. The ZnAl₂O₄ phase in the spinel structure enabled strong interaction between Al₂O₃ and ZnO [47]. As shown by the XRD, H₂-TPR, and TEM results, the Cu particles and ZnO in the matrix of the ordered mesoporous structure are small in size and show good dispersion, thereby increasing the desulfurization activities and enhancing sulfur capacity.

The calculated ZnO conversions of the different adsorbents at breakthrough points are presented in Table 2. The ordered-mesoporous-structure Cu-ZnO-Al₂O₃ adsorbents have higher ZnO conversions than the unordered mesoporous structure. In the RADS system, the S-compounds decomposed on the active metal surface, thereby sulfurizing it, followed by the transfer of sulfur to ZnO in the hydrogen condition [48]. ZnO acts not only as a support on the active metal, but also as an S-compound acceptor. The well-dispersed ZnO in the ordered mesoporous sample could facilitate the sulfidation of ZnO by H₂S with weak metal-support interaction [49]. This process is beneficial for maintaining the initial state of the active metal of Cu and improving ZnO conversion, thus sustaining good activity over long periods of the RADS process. Moreover, the ordered mesoporous structure could inhibit the string of Cu particles to improve the desulfurization activities and stability. Hence, compared with the unordered mesoporous structure, ordered mesoporous Cu-ZnO-Al₂O₃ adsorbents show good reactive activity, larger sulfur capacity, and stability.

In summary, a series of Cu-ZnO-Al₂O₃ adsorbents with ordered mesoporous structure were synthesized by the EISA strategy and were successfully applied for RADS. The ordered mesoporous structure is maintained even after the addition of 10 wt% CuO, for various ZnO contents between 0 and 20 wt%. The characterization results confirmed that for an appropriate amount of ZnO, the ordered mesoporous structure is maintained, resulting in thermal stability, and large specific surface areas and pore volumes in the Cu-ZnO-Al₂O₃ adsorbents. The ordered mesoporous Cu-ZnO-Al₂O₃ adsorbents exhibit larger sulfur saturation capacity than the unordered mesoporous adsorbent. The ordered mesoporous structure could provide a larger pore structure, thus enabling the high dispersion of ZnO; it is therefore useful for transforming S-compounds to ZnO to enhance the desulfurization activity. Thus, the ordered mesoporous structure of Cu-ZnO-Al₂O₃ adsorbents could enhance the desulfurization activity and prevent frequent regeneration during RADS.