Introduction

In the past decades, considerable attention has been paid to the supported gold catalysts for CO oxidation at low temperatures [1]. Various oxides, such as TiO2 [2], Fe2O3 [3], Al2O3 [4–6], CeO2 [7], MnOx [8] etc., have been employed to disperse and stabilize Au nanoparticles. Thereinto, gold catalysts supported on reducible metal oxides, in particular on CeO2, TiO2, Fe2O3, MnOx, are well known for their high activity for CO oxidation at low temperatures.

Among those nonreducible oxide supporting gold nanoparticles, Au/γ-Al2O3 is still a very interesting system for both practical application and academic study due to the extraordinary advantages of γ-Al2O3 such as high surface area and good mechanical and especially thermal, chemical stability [9]. However, the less active of Au/Al2O3 catalyst for CO oxidation has been observed experimentally. For example, gold nanoparticles supported on mesoporous γ-Al2O3 give a complete CO conversion when the reaction is performed at 150 °C [10]. When using commercial γ-Al2O3 support, the CO conversion of Au catalyst was only ca. 50 % at 65 °C (T 50%) and reached to 90 % at 100 °C. Even after Au nanoparticles supported on γ-Al2O3nanofibers, the enhanced catalytic activity can only achieve the complete CO conversion at 40 °C (T 100% = 40 °C) [11]. Thus, it remains a grand challenge to obtain highly active Au/Al2O3 catalyst for CO oxidation. Recently, we reported the γ-Al2O3with thin sheets and rough surface acted as an extraordinary catalyst support for the stabilization of gold nanoparticles from sintering [4]. It reveals that, apart from its intrinsic quality, the morphology of support material plays a significant role in the reactive activity of Au catalysts.

Herein, assembled from closely packed nanoflakes, γ-Al2O3 hollow microsphere structures with ca. 4–6 μm in diameter and 500–700 nm in shell thickness have been hydrothermally prepared through adjusting the molar ratio of Al3+ and SO4 2−, followed by a calcination step. As such unique structures of γ-Al2O3 support could efficiently stabilize Au nanoparticles, the obtained Au catalyst exhibits extraordinarily high activity towards CO oxidation, where a complete CO conversion at 0 °C was achieved (T 50% = −25 °C). This work could be distinguished by special hollow microspheres structures assembled by nanoflakes of γ-Al2O3 support and excellent catalytic activity for CO removal.

Experimental

Synthesis of Au/Al2O3 catalysts

Typically, aluminum nitrate, sulfates (K2SO4, (NH4)2SO4 or KAl(SO4)2·12H2O) and urea were dissolved in 100-mL deionized water. The obtained mixture was transferred into a 150-mL Teflon-lined stainless steel autoclave and heated at 180 °C for 3 h. Then, the white precipitates were washed with deionized water and dried at 80 °C. The hydrothermally synthesized and dried samples were denoted as Al 2 O 3 -x-hydro (where x = 1, 2, and 3, corresponding to different sulfate). After a calcination step at 500 °C for 2 h, the final products were obtained and named as Al 2 O 3 -x. The concentration of Al3+ was fixed as 0.05 M. The detail synthesis conditions and textural parameters were listed in Table 1. For comparison, alumina (denoted as Al 2 O 3 -4) was synthesized by a common precipitation method using aluminum nitrate as precursor and ammonium carbonate as precipitant.

Table 1 Synthesis conditions and structure parameters of γ-Al2O3 samples
Full size table

Preparation of Au/Al2O3 catalyst and catalytic test

Au nanoparticles were deposited on the surface of the sample Al 2 O 3 -x (x = 1, 2, 3, and 4) by a deposition–precipitation method with HAuCl4 solution (7.9 g L−1) at pH 8–9 in 60 °C for 2 h. After washing and drying, the precipitants were thermal treated at 250 °C for 2 h in the air to generate the Au/Al2O3 catalysts, denoted as Au-(Al 2 O 3 -x). The Au content was theoretically estimated as 3 wt.%. The activity of Au catalysts for CO oxidation was evaluated in a fixed bed quartz reactor using 50 mg of catalyst (20−40 mesh) with a composition of 1 vol.% CO, 20 vol.% O2, and 79 vol.% N2 and the total rate of feed gas was 67 mL min−1 (80,000 mL h−1 gcat −1). The products were analyzed using a GC-7890 gas chromatograph equipped with a thermal conductivity detector.

Characterization

Thermogravimetric and differential scanning calorimetry analysis (TG-DSC) were conducted on a thermogravimetric analyzer STA 449 F3 (NETZSCH) under an air atmosphere with a heating rate of 10 °C min−1. X-ray diffraction patterns (XRD) were obtained with a D/MAX-2400 diffractometer using Cu Kα radiation (40 kV, 100 mA, λ = 1.54056 Å). Nitrogen adsorption/desorption isotherms were measured with a TriStar 3000 adsorption analyzer (Micromeritics) at liquid nitrogen temperature. The samples were degassed at 200 °C for 4 h prior to analysis. The Brunauer–Emmett–Teller (BET) method was used to calculate the specific surface areas (S BET ). Pore size distributions (PSDs) were derived from the desorption branches of the isotherms using the Barrett–Joyner–Halenda model. Scanning electron microscope (SEM) images were obtained with a Hitachi S-4800 instrument. Transmission electron microscope (TEM) images were obtained with a Tecnai G220 S-Twin microscope with accelerative voltage of 200 kV. The remained K and S contents were analyzed using an inductively coupled plasma atomic emission spectrometer (ICP-AES) on the Optima 2000 DV. Infrared Fourier transform spectra (FT-IR) were recorded using a Nicolet 6700 FT-IR spectrometer at a resolution of 4 cm−1 and scale at 4,000–640 cm−1. Diffuse reflectance infrared Fourier transform spectra were recorded using a Nicolet 6700 FT-IR spectrometer at a resolution of 4 cm−1 and scale at 4,000–640 cm−1. Self-supporting disks were prepared from the sample powders and treated directly in the IR cell. The catalysts were connected to a vacuum-adsorption apparatus with a residual pressure below 10−3 Pa. Prior to the CO adsorption (25 °C, 5 vol.% CO and N2 in balance), the catalysts were evacuated for 30 min for 1 h at 100 °C.

Results and discussion

Alumina hollow microspheres

To explore the thermal decomposition behavior, hydrothermal products were characterized by TG-DSC (Fig. 1a, b) under air, with a heating rate of 10 °C min−1. In the case of samples Al 2 O 3 -x-hydro (x = 1, 2, and 3), the TG curves show a sharp weight loss before 470 °C with the actual loss is ~18 %, mainly corresponding the release of adsorbed water and the decomposition of AlO(OH) (2AlO(OH) → Al2O3 + H2O, theory weight loss value is ~15 %) [12]. The DSC curve displays a broad exothermic peak at 420 °C, which indicates the phase conversion. We thus selected 500 °C as the calcination temperature. However, Al2O3-4-hydro sample shows the weight loss of ~34 % below 600 °C, which is corresponding to the intermediate product Al(OH)3 (2Al(OH)3 → Al2O3 + 3H2O, theory weight loss value is ~34.6 %) [13].

Fig. 1
figure 1

a TG and b DSC curves of hydrothermal products (without calcination), c XRD patterns and d FT-IR curves of alumina supports (calcined at 500 °C for 2 h)

Full size image

Based on the XRD analysis results in Fig. 1c, all the diffraction peaks of calcined samples Al 2 O 3 -x (x = 1, 2, 3, and 4) can be assigned to γ-Al2O3 (JCPDS No. 10-0425). According to the Scherrer equation, the crystallite size of nanoflake composed hollow microsphere is ca. 13, 10, 12, and 10 nm for Al 2 O 3 -1, Al 2 O 3 -2, Al 2 O 3 -3, and Al 2 O 3 -4, respectively. More characteristics were detected in the FT-IR spectra (Fig. 1d). Basically, four samples show similar IR signals. The intensive bands at 3,447–3,449 and 1,637–1,638 cm−1 belong to the stretching and bending vibrations of the O–H bonds adjoining the Al atoms and a functional group of water [14–16]. The intensive peaks at 1,160 and 1,070 cm−1 are due to the δ as Al-O-H and δ s Al-O-H modes, and the bands observed at 746 cm−1 represent the stretching modes of AlO6 [17].

After annealing at 500 °C for 2 h, the alumina materials were further characterized by N2 sorption measurements. As seen in Fig. 2, the nitrogen sorption isotherms of Al 2 O 3 -1, Al 2 O 3 -2, Al 2 O 3 -3, and Al 2 O 3 -4 are essentially of type IV [18], reflecting a mesoporous characteristic. Determined from desorption branches, the pore size distributions of Al 2 O 3 -1, Al 2 O 3 -2, and Al 2 O 3 -3 are centered at 4/17, 4/17, and 3 nm, whereas Al 2 O 3 -4 shows a rather broad distribution with mesopore size concentrated at 19 nm. Al 2 O 3 -4 exhibits a slightly higher BET surface area (268 m2 g−1) and total pore volume (0.92 m3 g−1) than those of Al 2 O 3 -3 (209 and 0.66 m3 g−1, accordingly). The detail structure parameters of γ-Al2O3 materials are listed in Table 1.

Fig. 2
figure 2

a N2 sorption isotherms and b pore size distributions of Al 2 O 3 -1, Al 2 O 3 -2, Al 2 O 3 -3 and Al 2 O 3 -4 aluminas. The isotherms of Al 2 O 3 -4, Al 2 O 3 -1 and Al 2 O 3 -2 were offset up by 400, 300, and 100 cm3 g−1

Full size image

In addition, the morphologies of alumina were characterized by SEM technique. It can be clearly seen in Fig. 3a, b that Al 2 O 3 -1 and Al 2 O 3 -2 samples display hollow microsphere structures (the diameter ~4 μm and shell thickness 500–600 nm) with open mouths assembled from closely packed nanoflakes. When KAl(SO4)2·12H2O was used as alumina precursor and sulfate additive (keeping the molar ratio of Al:SO4 2− = 1:2), then, the perfect γ-Al2O3 hollow microspheres with ca. 6 μm in diameter, 600–700 nm in shell thickness can be obtained (Fig. 3c). Whereas, Al 2 O 3 -4 has an undefined morphology (Fig. 3d), which is totally different from alumina hollow microspheres prepared with the addition of sulfates. As the promotion of SO4 2−, Al3+ and urea can alternatively hydrolyze and polycondense, which leads to the precipitation of amorphous aluminum oxyhydroxide spheres [18–23]. It can be deduced that the increase in concentration of SO4 2− facilitates formation of larger microspheres through secondary nucleation and growth of AlO(OH) crystal nuclei until the system reaches equilibrium with the surrounding solution. It confirms that the sulfate anion plays a significant role in the formation of hollow microspheres morphology.

Fig. 3
figure 3

SEM images of alumina supports (a Al 2 O 3 -1, b Al 2 O 3 -2, c Al 2 O 3 -3, d Al 2 O 3 -4)

Full size image

Au/Al2O3 catalyst for CO oxidation

As a simple and typical probe reaction, CO oxidation was selected to identify the promotion of γ-Al2O3 with unique hollow microspheres morphology assembled by closely packed nanoflakes for the catalytic performance of Au nanoparticle catalysts. Supported on obtained γ-Al2O3 hollow microspheres, the catalytic activity of 3 wt.% Au/Al2O3 catalysts were shown in Fig. 4 and Table 2.

Fig. 4
figure 4

CO conversion curves of gold catalysts Au-(Al 2 O 3 -1), Au-(Al 2 O 3 -2), Au-(Al 2 O 3 -3) and Au-(Al 2 O 3 -4)

Full size image
Table 2 Catalytic activity of gold nanoparticles supported on different alumina
Full size table

Remarkably, the corresponding Au catalysts exhibited excellent catalytic performance with a complete CO conversion at ~0 °C. Particularly in the case of sample Au-(Al 2 O 3 -3), the 50 % CO conversion can be achieved at −25 °C, and the specific rate at 0 °C was calculated as 1.413 mol h−1 gAu −1. For comparison, the regular Al2O3 support (Al 2 O 3 -4) with undefined morphology was synthesized by a common precipitation method using Al(NO3)3·9H2O and (NH3)4CO3. The obtained Au-(Al 2 O 3 -4) catalyst tested under the same conditions is generally inactive at 0 °C. The results in turn indicate that such γ-Al2O3 hollow microspheres can efficiently stabilize Au nanoparticles, which exhibit higher catalytic performance for CO oxidation. In addition, despite the different catalytic test conditions, such as the composition of feed gas, the loading content of gold particles, and the temperature of calcination, the rate values in this work are comparable and even higher to the results in literatures [10, 13, 24–27].

Presently, from SEM images (Fig. 3) we know Al2O3-1, Al2O3-2, and Al2O3-3 have quite similar morphologies while Al2O3-4, as controlled sample, is in another case. On the other hand, the activity of Au/Al2O3-1, Au/Al2O3-2, and Au/Al2O3-3 samples are also quite similar in CO oxidation, while Au/Al2O3-4 shows a very low activity (Fig. 4). The Au contents of all samples were maintained nearly same (Table 2). Therefore, one representative sample Au/Al2O3-3 was characterized by TEM. As shown in Fig. 5a, b, gold catalysts supported on the surface of γ-Al2O3 hollow microspheres consisted of closely packed nanoflakes. Besides, Au nanoparticles are highly dispersed with an average size of 3.0 ± 0.5 nm estimated from the TEM images in Fig. 5c. In addition, Fig. 5d shows a rough surface of γ-Al2O3 hollow microspheres, which are beneficial to stabilize gold nanoparticles efficiently. From our previous work [13], the controlled sample Au/Al2O3-4 has a gold particle size of 2.5 ± 0.5 nm based on the TEM observation, which is more or less similar to that of sample Au/Al2O3-3. All these information demonstrates the advantages of the rough surface of γ-Al2O3 hollow microspheres. Furthermore, concerning the influence of the possible remained heteroatoms (for example, sulfur and potassium) on the activity [28–30], sulfur contents and potassium contents of samples Al 2 O 3 -1, Al 2 O 3 -2 and Al 2 O 3 -3 were analyzed by ICP-AES technique, correspondingly. All three samples show ignorable sulfur and potassium content. Therefore, one can basically eliminate the influence of sulfate and K+ on the catalytic performance. It further suggests that the shape and surface property of γ-Al2O3 support have strong influence on the activity of the Au nanoparticles.

Fig. 5
figure 5

TEM images of gold catalyst Au-(Al 2 O 3 -3) and size distribution of gold nanoparticles

Full size image

To further explore the surface chemical property of Au nanoparticles on γ-Al2O3 hollow microsphere supports, in situ DRIFTS was applied to measure the adsorbed CO. As shown in Fig. 6, the strong absorption bands of CO positioned around 2,110 and 2,070 cm−1 appeared on the Au/Al2O3 samples dispersed on γ-Al2O3 hollow microspheres which belong to two linear CO species adsorbed on the metallic Au sites [31]. The peak at 2,170 cm−1 was attributed to CO adsorbed on cationic gold (Au(I) or Au(III)) [32]. The DRIFTS measurements suggest that the presence of Auδ+ and Au0 on the surface of Au/Al2O3 catalysts, although weaker absorption bands were observed for Au nanoparticles supported on alumina Al 2 O 3 -4. Meanwhile, the absorption band appeared at 2,052 cm−1, which can be assigned to CO adsorbed on a negatively charged gold surface [33]. Apparently, Al 2 O 3 -4 synthesized by the common precipitation method displays an inferior surface for CO adsorption and thus results in a catalyst with a low catalytic activity.

Fig. 6
figure 6

In situ DRIFTS spectra of gold catalysts Au-(Al 2 O 3 -1), Au-(Al 2 O 3 -2), Au-(Al 2 O 3 -3) and Au-(Al 2 O 3 -4)

Full size image

Conclusions

It has been demonstrated that AlO(OH) hollow microspheres consist of closely packed nanoflakes were synthesized through a hydrothermal process by using Al(NO3)3·9H2O as a precursor, urea as precipitant agent and sulfate K2SO4, (NH4)2SO4, and KAl(SO4)2·12H2O as additive. When adjusting the molar ratio of Al3+ and SO4 2− to 1:2, γ-Al2O3 hollow microspheres with ca. 4–6 μm in diameter, 500–700 nm in shell thickness can be obtained after a calcination step. As such, γ-Al2O3 supports with special morphology and rough surface can stabilize gold nanoparticles efficiently; an excellent catalytic performance for CO oxidation can be achieved with a complete CO conversion at 0 °C and 50 % conversion at −25 °C. The DRIFTS confirms the presence of Auδ+ and Au0 on the surface of Au/Al2O3 catalysts. In contrast, Au catalyst on alumina support with undefined morphology shows low activity at low temperature under the same catalytic test conditions.