Salt-induced formation of hollow and mesoporous CoO_x/SiO₂ spheres and their catalytic behavior in toluene oxidation

Abstract

Hollow and mesoporous CoO_x/SiO₂ spheres (denoted as CoO_x/hSiO₂ and CoO_x/mSiO₂) were synthesized via salt-assisted ultrasonic spray pyrolysis. Precursor solutions containing sodium silicate solution, a mineral acid (hydrochloric acid or nitric acid) and a cobalt salt were ultrasonically aerosolized and pyrolyzed. Results showed that CoO_x/SiO₂ spheres with hollow or mesoporous structure can be fabricated by using the NaCl and NaNO₃ salts as in situ formed templates. Significantly, this approach avoids the need for post-calcination for template elimination, instead permitting aqueous removal with water. The influence of the sodium salts on the characteristics of CoO_x/SiO₂ spheres was investigated by means of XRD, nitrogen physisorption, SEM/TEM, ICP-MS, UV-Vis, XPS and H₂-TPR. On the basis of the experimental results, a possible mechanism on the formation of hollow and mesoporous spheres was proposed. The CoO_x/SiO₂ spheres were tested as catalysts for toluene oxidation. The mesoporous CoO_x/mSiO₂ was found to exhibit superior activity to hollow CoO_x/hSiO₂, probably attributed to a combination of several factors, including the predominant existence of a Co₃O₄ active phase, high surface Co³⁺ content, and easy reducibility of Co³⁺ at low temperature.

---

1. Introduction

Catalytic oxidation has been proposed as an effective and energy-saving technology for eliminating volatile organic compounds (VOCs) emitted from industrial processes and transport vehicles.^1–3 Recently, cobalt oxide (Co₃O₄) has attracted much attention to control the emissions of VOCs, and it has been considered as a promising alternative to costly noble metal-based catalysts because of its high activity, relatively low price and availability.^4,5 However, the insufficient catalytic performance at low temperatures, narrow operating-temperature window and poor thermal stability of Co₃O₄ largely limit its performance and impede the commercialization of VOC removal.^6–9

Several studies have been conducted to improve the catalytic activity of cobalt oxide by depositing it on various supports, such as Al₂O₃, ZrO₂, TiO₂, SiO₂ and CeO₂, etc.^10,11 Among available support materials, ordered mesoporous silicas synthesized by surfactant-templating methods have been investigated as promising candidates for catalysis because of their own unique characteristics of high surface area and regular pore structure, which can be used to uniformly disperse the active metal/metal oxide particles.^12–15 Numerous mesoporous silicas, such as MCM-41, SBA-15 and KIT-6 loaded with cobalt oxides for VOCs oxidation have been reported.^16–20 The performance of the supported cobalt oxide catalyst is highly dependent on different synthetic methods as well as precursors.^21,22 Moreover, the crystallinity, oxidation state and surface reducibility of supported cobalt oxide species have been shown to be critical to the activity of the catalyst.^23,24 For preparing highly active catalysts in VOCs oxidation, it is important to obtain Co₃O₄ oxide phase with high surface reducibility.

Although these novel cobalt oxide/mesoporous silica composite catalysts have contributed to recent progress in VOCs removals, they may pose problems in practical field applications because of their high cost.²⁵ Generally, these composites are synthesized through solution-based batch processes that take several days in a high pressure vessel and require further post-treatments, such as solvent extraction and calcination for template removals.^22,26 Therefore, they are difficult to apply to commercial scale production due to their long processing times, significant batch-to-batch variation and high energy consumption.

Compared to solution-based methods, the aerosol-assisted chemical vapor decomposition, also known as spray pyrolysis, has distinct advantages of rapid synthesis, high product purity, and more importantly, a continuous and scalable process.^27–31 This process involves atomization of a liquid precursor into fine aerosol droplets that are delivered to a heated zone where solvent evaporation and precursor decomposition take place, leading to the deposition of final solid product. The detailed descriptions on this technique can be found in several reviews.^31–33 Recently, Debecker and co-workers³⁴ synthesized a series of mesoporous WO₃–SiO₂–Al₂O₃ microspheres by the approach combining spray pyrolysis and evaporation induced self-assembly (EISA) of poly(alkylene oxide) block copolymer. Nevertheless, most of the aerosol-based methods reported previously were carried out using expensive silicon precursors (e.g. silicon chloride and silicon alkoxide) and sacrificial templating agents (e.g. surfactant polymers and colloidal polystyrene) as starting precursors,³⁵ which would increase the manufacturing cost and add pollution to the environment.

In this study, we first report the salt-assisted ultrasonic spray pyrolysis fabrication of cobalt oxide/silica hollow and mesoporous spheres (detonated as CoO_x/hSiO₂ and CoO_x/mSiO₂) using sodium silicate and cobalt nitrate hydrate as precursors. Compared with traditional aerosol-assisted EISA method for preparing mesoporous silica supported catalysts where costly precursors are employed, the synthesis strategy demonstrated herein is much more inexpensive and energy-saving because no extra templating agents, or high-temperature calcination are required for template removals. The influence of salts on the characteristics of produced spheres is studied in detail by various characterization tools. On the basis of the experimental results, the formation mechanism of hollow and mesoporous spheres is proposed. Further, our experimental results reveal that, compared with pristine Co₃O₄, the CoO_x/mSiO₂ demonstrates greatly improved catalytic performance and long-term thermal stability for toluene removal.

2. Experimental section

2.1 Catalyst preparation

Hollow and mesoporous CoO_x/SiO₂ composites were continuously synthesized by a single-step aerosol process in a home-made apparatus depicted in Scheme 1. The sodium silicate solution, which is recovered from the silicate waste of liquid crystal display (LCD) manufacturing industry, is used as a low-cost silicon source in this study.^36,37 First, the sodium silicate solution was prepared by mixing LCD industrial waste powder with 6 M NaOH solution at room temperature for 3 h. The elemental composition of raw waste powder and silicate supernatant after extraction with NaOH solution are summarized in Table S1.† Thereafter, a certain amount of HCl or HNO₃ was added to the 189 ml of waste silicate supernatant to bring down the pH value to 2 with constant stirring. Meantime, a calculated amount of cobalt nitrate (dissolved in 20 ml of DI water) was added into the above acidified silicate solution and the combined mixture was stirred for 30 min. The molar composition of the precursor mixture was 1 SiO₂ : 2 Na : 0.055 Co : 280 H₂O : 8 HCl, and 1 SiO₂ : 2 Na : 0.055 Co : 280 H₂O : 3.7 HNO₃, for CoO_x/hSiO₂ and CoO_x/mSiO₂, respectively.

Scheme 1 Schematic drawing illustration the aerosol route and the washing process of CoO_x/hSiO₂ and CoO_x/mSiO₂ particles.

The precursor mixture was nebulized by an ultrasonic atomizer (1.8 MHz) as carried by high-pressure air (35 psi). The temperature of the heating reactor was controlled at 500 °C, and the total synthesis time of this continuous flow process was approximately 5 s to generate the CoO_x/SiO₂ particle. The as-synthesized material was collected downstream of the reactor with a high efficiency filter. Finally, they were recovered by washing and filtration with DI water followed by drying in an oven at 110 °C.

To verify the activities of CoO_x/SiO₂ catalysts, the aerosol-made CoO_x/HMSP (hexagonally mesoporous silica particle), which was prepared using commercial silicon precursor via traditional surfactant-templated method, was prepared with similar procedures described in the synthesis of CoO_x/SiO₂ materials. Tetraethoxysilane (TEOS) and cetyltrimethylammonium bromide (CTAB) was selected as the silicon source and mesostructure-directing template, respectively. The molar gel composition of the mixture was 1 SiO₂ : 0.18 CTAB : 0.055 Co : 10 ethanol : 80 H₂O : 0.008 HCl. The as-prepared CoO_x/HMSP powder was then placed in a muffle furnace for 6 h at a calcination temperature of 550 °C to remove the residual organic template. Detailed description on the synthesized procedure can be referred to Wang and Bai.^38,39 The cobalt loading on the silica support for all supported catalysts was ca. 5 wt% according to our experimental results.

The pure Co₃O₄ used in this work was prepared by calcination of Co(NO₃)₂·4H₂O at 500 °C for 4 h.

2.2 Characterization

The chemical composition of raw waste powder was determined by energy-dispersive X-ray spectroscopy in a scanning electron microscopy (SEM, HITACHI-S4700). The elemental cobalt content in the samples was analyzed by a SCIEX ELAN 5000 inductively coupled plasma-mass spectrometer (ICP-MS). The morphology and structure of the samples were characterized by using SEM (HITACHI-S4700), transmission electron microscopy (TEM, Hitachi H-7100, Japan) and X-ray diffraction (XRD, Rigaku D/MAX-B, Japan, using Cu Kα radiation n source at 1.54 Ǻ). Nitrogen physisorption isotherms of the synthesized materials were measured at −196 °C using a surface area and pore diameter distribution analyzer (Micromeritics, ASAP 2020, USA). Pore volumes were obtained from the volumes of nitrogen adsorbed at P/P_o = 0.95 or in the vicinity.

The UV-Vis spectra were measured by a spectrophotometer (HITACHI U3010) equipped with a diffuse reflectance integrating sphere coated with aluminum oxide which served as a reference material. All the samples were recorded in the spectral range between 900 and 275 nm. X-ray photoelectron spectroscopy (XPS) measurement was performed on a PHI Quantera with a monochromatic Al Kα source and a charge neutralizer. All the binding energies were calibrated to C 1s peak at 284.8 eV of the surface adventitious carbon. H₂-TPR experiments were performed with an AutoChem II 2920 analyzer. The samples were pre-treated in air at 500 °C for 2 h. After that, the experiments were carried out from 50 °C to 900 °C at a heating rate of 10 °C min⁻¹ in 10% H₂/Ar. The H₂ consumption was determined by a thermal conductivity detector (TCD).

2.3 Catalytic removal of gaseous toluene

Catalytic tests were performed by a vertical and downward flow reactor system (i.d. = 0.8 cm). Catalysts were tested in 16–30 mesh (595–1190 μm) powdered form, and a total of 0.1 g catalyst was packed into the middle of the glass reactor supported with thin layers of glass wool on both sides. The reactant feed was composed of 1000 ppm toluene, O₂ and N₂ (balance). The toluene/O₂ molar ratio was 1/200, and the total inlet flow rate was controlled at 100 cm³ min⁻¹, which corresponded to a space velocity (SV) of 20 000 h⁻¹ at room temperature (25 °C). Prior to each catalytic reaction test, the samples were pretreated under the reaction feed by heating at 120 °C for 2 h, and then cooled to room temperature. Then, reaction temperature was increased stepwise from 100 °C to 500 °C. The catalytic activity was evaluated in terms of toluene conversion. When the complete conversion for toluene oxidation was obtained, the temperature was kept for 33 h to test the reaction stability of the catalyst. To evaluate the reusability of the catalyst, the reuse runs were implemented after the stability test. Each run was kept at 250 °C for 2 h, then the reaction was cooled down to the room temperature. The gas composition after the reaction was analyzed by a gas chromatograph equipped with a flame ionization detector (FID) and a TCD detector.

3. Results and discussion

3.1 Catalyst characterization

The wide-angle (10 ≦ 2θ ≦ 80°) powder XRD patterns of as-prepared CoO_x/hSiO₂ and as-prepared CoO_x/mSiO₂ are depicted in Fig. 1A. It can be seen that several diffraction reflections are found in both as-prepared CoO_x/hSiO₂ and CoO_x/mSiO₂, which can be indexed on crystalline NaCl and NaNO₃, respectively. Such an observation can be explained by considering the acidification reaction of sodium silicate by HCl or HNO₃. In the acidification reaction, the silica condensation reactions would occur to form a siloxane linkage between surface silanol groups, which can be represented as:⁴⁰

Si(OH)₄ + HO–Si(OH)₃ → (OH)₃Si–O–Si(OH)₃ + H₂O (1)

Fig. 1 (A) Wide-angle XRD patterns of as-prepared CoO_x/hSiO₂ and as-prepared CoO_x/mSiO₂; (B) wide-angle and (C) low-angle XRD patterns of washed CoO_x/hSiO₂, washed CoO_x/mSiO₂, CoO_x/HMSP and Co₃O₄.

And the acidification reaction of sodium silicate solution with HCl and HNO₃, respectively, can be expressed as:

Na₂O·xSiO₂ + 2HCl → xSiO₂↓ + 2NaCl + H₂O (2)

Na₂O·xSiO₂ + 2HNO₃ → xSiO₂↓ + 2NaNO₃ + H₂O (3)

At this time, soluble sodium salts of NaCl or NaNO₃ were in situ formed. During the aerosol process (T = 500 °C), the rapid water evaporation would result in the solidification of these sodium salts, which were then embedded in SiO₂ spheres.

For washed CoO_x/SiO₂ samples (Fig. 1B), there is a broad reflection at 23° observed for both two samples, indicating that silica materials are presented in amorphous phase.⁴¹ The lack of visible diffraction reflections due to crystalline NaCl or NaNO₃ components reveals that these sodium salts can be easily removed by washing with water. Moreover, it is noticeable that three weak diffraction reflections at 2θ = 36.9, 59.4 and 65.2° are found in CoO_x/mSiO₂. These diffraction reflections can be assigned to the spinel-structured Co₃O₄.²² The presence of visible Co₃O₄ reflections reveals that the Co₃O₄ particles in CoO_x/mSiO₂ might be presented in larger particle size or in agglomerated form. On the contrary, both CoO_x/hSiO₂ and CoO_x/HMSP show no observable reflections of Co₃O₄. This could be attributed to that the small-sized Co₃O₄ are uniformly distributed and the size of Co₃O₄ particles are below the detection limit of the XRD instrument.⁴² The low-angle (1 ≦ 2θ ≦ 10°) powder XRD pattern (Fig. 1C) of CoO_x/HMSP shows two well-defined diffraction reflections of (100) and (110) at 2θ of 2.4° and 4.3°, revealing the well-ordered hexagonal mesostructure of CoO_x/HMSP.³⁹

The particle morphologies of CoO_x/SiO₂ materials are revealed by SEM and TEM images. As seen in Fig. 2A and B that CoO_x/hSiO₂ consists of mostly hollow shells or fractured hollow shells (while remaining the spherical shape). In addition, many of these shell structures contain comparatively rough and porous surface. In contrast, as illustrated in Fig. 2D and E, spherical particles with well-defined morphologies are observed on CoO_x/mSiO₂. It is noteworthy that some well-dispersed tiny particles are locating on the external surface or sitting inside the hollow spheres (Fig. 2C). The tiny particles could be assigned to the presence of cobalt oxide particles. And it is found from the HRTEM images (Fig. 2F) that the cobalt oxide particles mainly exist between the interconnected pores of CoO_x/mSiO₂.

Fig. 2 SEM, TEM and selected area electron diffraction images of CoO_x/hSiO₂ (A, B and C) and CoO_x/mSiO_2- (D, E and F).

Selected area electron diffraction (SAED) image (Fig. 2F, inset) confirms the crystalline nature of the large Co₃O₄ particles (CoO_x/mSiO₂ sample). By contrast, no clear diffraction pattern was observed for the CoO_x/hSiO₂ sample, suggesting that the cobalt oxides are poorly crystallized. These findings are consistent with the XRD measurements (Fig. 1A). The absence of diffraction reflections indicates that the crystalline domains are very small (at most a few nanometers) and points to polycrystallinity. The CoO_x/hSiO₂ and CoO_x/mSiO₂ were further characterized through mapping analysis to investigate the Co distribution. As shown in Fig. S2,† it can be seen that Co species are well-dispersed in both hSiO₂ and mSiO₂ matrices. On the basis of the above observations, it may conclude that the nature of sodium salt has significant impacts not only on the morphology, but also on the crystallinity of the cobalt oxide particle of CoO_x/SiO₂ composites.

The pore structures of the prepared materials are further analyzed by the nitrogen physisorption measurement, with results displayed in Fig. 3A. The CoO_x/hSiO₂ exhibits type IV isotherms that possess two-step adsorption isotherm. This type of hysteresis loop has been reported to be associated with hollow particles with nanoporous shell.⁴³ An adsorption at P/P_o values in the range of 0.3–0.7 could be related to the capillary condensation of nitrogen inside nanopores in the shells, while the second adsorption at P/P_o = 0.7–0.9 corresponded to the filling of huge hollow cores, as observed in Fig. 2A and B. For CoO_x/mSiO₂ sample, a typical single-step type IV isotherms with H2-typed hysteresis loop is observed, which is indicative of mesoporous material having networks of interconnected pores.^44,45 Similarly, CoO_x/HMSP shows a single-step type IV isotherms with clear hysteresis loops at a relative pressure of P/P_o = 0.25–0.45, which is attributed to the presence of intra-particle mesoporosity originated from the surfactant template.^46,47

Fig. 3 (A) Nitrogen physisorption isotherm, (B) UV-Vis spectra, (C) XPS spectra and (D) H₂-TPR profiles of CoO_x/hSiO₂, CoO_x/mSiO₂, and CoO_x/HMSP.

The pore size distributions of CoO_x/hSiO₂ and CoO_x/mSiO₂ calculated by using NLDFT method are shown in Fig. S1,† and their physico-chemical parameters derived from nitrogen physisorption measurement are summarized in Table 1. It is observed that the formation of multi-modal porous particles are obtained for both CoO_x/hSiO₂ and CoO_x/mSiO₂. Such phenomenon presumably results from nonuniform solute distribution within the droplet.⁴⁸ Compared with CoO_x/SiO₂ samples, the CoO_x/HMSP, which was prepared through surfactant-templating route, clearly exhibits a narrow pore size distribution, suggesting the existence of uniform mesoporosity. And it also exhibits the highest surface area and largest pore volume among all catalysts studied (Table 1).

Table 1 Physicochemical parameters of cobalt-based samples

Sample name	SBET (m^2 g^−1)	Vpore (cm^3 g^−1)	Na content (wt%)	Cobalt content (wt%)	Binding energy (eV) Δ(Co 2p3/2–2p1/2)
As-prepared CoOx/hSiO2	13	0.04	7.35	—	—
As-prepared CoOx/mSiO2	20	0.08	7.48	—	—
Washed CoOx/hSiO2	411	0.34	0.42	3.09	15.8 (796.7–780.9)
Washed CoOx/mSiO2	471	0.47	0.44	3.95	15.2 (794.8–779.6)
Calcined CoOx/HMSP	872	0.61	—	4.88	16.3 (797.5–781.2)
Co3O4	13	0.03	—	—	—

---

The effect of sodium salts on the physico-chemical properties is also investigated and the results derived from nitrogen physisorption measurement and ICP-MS analyses are summarized in Table 1. It is seen that both as-prepared CoO_x/SiO₂ samples show high Na contents, low specific surface area and no porosities. After washing process with water, the surface area and porosities of both samples are significantly increased; meanwhile, the Na content in the as-prepared CoO_x/hSiO₂ and CoO_x/mSiO₂ samples are reduced to 0.42 and 0.44 wt%, respectively, in the washed CoO_x/SiO₂ samples. The above results clearly reveal that the NaCl and NaNO₃ salts, which are in situ formed during the acidification process of sodium silicate, can act as effective porogen, which can be then readily removed by washing with water. On the other hand, the cobalt content in the CoO_x/hSiO₂ and CoO_x/mSiO₂ is 3.09 and 3.95 wt%, respectively, which is lower than the theoretical content (5 wt%) in the precursor solution. This is probably due to the incomplete decomposition of cobalt precipitates that would result in some of the cobalt content being carried away during the washing process. Thus, one can conclude that the sodium salts have significant influences not only on the textural properties, but also on the chemical composition of the CoO_x/SiO₂ samples.

The UV-Vis spectra of the supported cobalt catalysts are shown in Fig. 3B. For CoO_x/hSiO₂ and CoO_x/mSiO₂ samples, two absorption bands at 370–390 and around 700 nm are found, which can be related to the Co³⁺ in the octahedral coordination in the Co₃O₄ and to the Co²⁺ in the tetrahedral coordination in the Co₃O₄, indicating the presence of spinel-structured Co₃O₄.^24,49 For CoO_x/HMSP, the wide bands at 525, 585 and 640 nm, typical for tetrahedrally coordinated Co²⁺ due to the metal–ligand charge transfer, are observed. They could be included in amorphous cobalt oxide clusters and Co²⁺–silicate species.²⁴ Thus, the spectra indicate the formation of various cobalt oxide species, which are in different interaction with the support.

The XPS analyses were conducted to determine the oxidation state of surface cobalt species, with results shown in Fig. 3C and Table 1. In the Co XPS spectra, the peaks located at 794.8–797.5 and 779.6–781.2 eV can be ascribed to Co 2p_1/2 and Co 2p_3/2 spin–orbital peaks, respectively.²² It can be seen the binding energy of Co 2p_3/2 for CoO_x/mSiO₂ is 779.6 eV, and the spin–orbit separation between Co 2p_3/2 and Co 2p_1/2, Δ(Co 2p_3/2–2p_1/2) of CoO_x/mSiO₂ is 15.2 eV (Table 1). The above results are indicative of the existence of Co³⁺ in Co₃O₄.²² For CoO_x/hSiO₂ and CoO_x/HMSP samples, the binding energies of Co 2p_3/2 are increased to 780.9 and 781.2 eV, respectively, accompanying with intense satellite peaks at ca. 785.5 eV, which have been reported as the evidence of the progressive formation of Co²⁺ species. Meanwhile, the Δ(Co 2p_3/2–2p_1/2) of CoO_x/hSiO₂ and CoO_x/HMSP increase to 15.8 and 16.3, respectively, demonstrating the decrease in the surface Co³⁺ content.^22,50 Besides, the relative surface contents of Co³⁺ (Co³⁺/(Co³⁺+ Co²⁺)), which was calculated from peak I/peal II area in Fig. 3C, decreases in the order of CoO_x/mSiO₂ > CoO_x/hSiO₂ > CoO_x/HMSP. Accordingly, one can conclude that the synthetic procedure greatly influences the oxidation state of surface cobalt species. The surface Co³⁺ content is the highest on CoO_x/mSiO₂, while CoO_x/HMSP exhibits the lowest content of surface Co³⁺.

The surface reducibility of the catalysts studied by the H₂-TPR measurement are displayed in Fig. 3D. For CoO_x/mSiO₂ sample, four main reduction peaks centering at 279, 482, 720 and 771 °C can be clearly observed. By analogy with previous studies, the first peak is indicative for the reduction of Co³⁺ to Co²⁺, while the second peak is associated with the reduction of CoO to metallic Co.^20,21,50 The broad peak between 600 to 900 °C is attributed to the reduction of the remaining Co²⁺ and highly dispersed tetrahedral coordinated Co²⁺–silicate-like species with strong interaction with the support.^51,52 Compared with CoO_x/mSiO₂, CoO_x/hSiO₂ sample displays a progressive shift of the H₂ consumption peak to higher temperatures, indicating decreased reducibility. The TPR profile of CoO_x/HMSP differs significantly from those of other samples. Three weak reduction peaks can be found at 262, 380, and 535 °C, respectively, which could be ascribed to the reduction of Co³⁺ and CoO to metallic Co. Additionally, a sharp reduction peak at 730 °C with a shoulder at 806 °C are also observed. They might be mainly originated from the reduction of Co²⁺ and Co²⁺–silicate-like species, and the above observation is in good agreement with UV-Vis and XPS results. It has been reported that the nature of the support affects the particle size, and the low-surface-area supports promote the agglomeration of cobalt particles as larger Co₃O₄ crystals, which are easily reduced in comparison with smaller particles.^23,49 According to this, the high-surface-area HMSP support may provide a suitable environment for the formation of small-sized and highly dispersed cobalt oxide species, which interact strongly with the support, and thus had a detrimental effect on the surface reducibility.

It has been reported that reduction of Co³⁺ species is one of the crucial factors that determines the activity of cobalt oxide catalyst.^22,53 Wu et al.²¹ showed that Co₃O₄–CeO₂ mixed oxides with a lower reduction temperature of Co³⁺ presented higher catalytic activity in methane oxidation. It is clear that high surface reducibility favors high catalytic activity since the reduction of reactive surface species frequently occurs at relatively lower temperature. In this study, for the reduction of Co³⁺ in the TPR profiles, CoO_x/mSiO₂ has a relatively lower reduction temperature (279 °C) and the highest H₂ consumption of Co³⁺ amount among all supported catalysts. This indicates that CoO_x/mSiO₂ has the highest amount of easily reducible Co³⁺. On the other hand, in spite of the lowest reduction temperature of Co³⁺ (262 °C), CoO_x/HMSP shows the lowest H₂ consumption amount of Co³⁺, suggesting that it contains the least amount of easily reducible Co³⁺ species.

3.2 Formation mechanism

It is of particular interest to understand the formation mechanism of CoO_x/hSiO₂ and CoO_x/mSiO₂ particles considering that the hollow shell and mesoporous morphologies are obtained without the use of extra templates. On the basis of the aforementioned results, a possible formation pathway of CoO_x/hSiO₂ and CoO_x/mSiO₂ particles is proposed in Scheme 2. At first, precursor sols were firstly prepared via the acidification of LCD waste-derived sodium silicate solution by adding either HCl or HNO₃; meanwhile, the soluble sodium salts of NaCl or NaNO₃ were in situ formed. Subsequently, cobalt nitrate was added as a metal precursor.

Scheme 2 Scheme representation showing the formation mechanism of CoO_x/hSiO₂ and CoO_x/mSiO₂ through salt-assisted aerosol process.

As the precursor was aerosolized and entered a heated zone (500 °C), the rapid water evaporation resulted in a temperature gradient within the aerosol droplet, with the surface of the aerosol droplets being heated first and then heat being transferred to the core. Secondly, when heating of the droplet proceeded, water containing dissolved salts like NaCl diffused from the interface to the core due to the improved ion concentration at the interface because of water evaporation; simultaneously, the silica and cobalt species would cross-link with each other and form cobalt–silicate species by isolating the structural hydroxyl neighbors during the thermal treatment at 500 °C. The cobalt–silicate aggregate acted as a shell, and salts began to nucleate and be fixed by primary units. Thirdly, when water evaporated completely, NaCl crystallized and aggregated as a core, while the thermal decomposition of the precursor of cobalt–silicate might yield cobalt hydroxide, which was subsequently transformed to CoO/Co₃O₄. Finally, CoO_x/hSiO₂ hollow particles were produced after the removal of NaCl by washing with water.

In contrast to CoO_x/hSiO₂, CoO_x/mSiO₂ particle with well-defined morphology and mesoporous structure is observed. Similarly, the NaNO₃ diffused from the interface to the core due to rapid water evaporation during aerosol process. As the process continued, the NaNO₃ (melting point = 308 °C) would become molten and served as a solvent, diffusing back into the interface in response to the concentration gradient. After washing with water, CoO_x/mSiO₂ having interconnected mesopores were obtained.

Moreover, unlike NaCl, which is thermally stable up to 800 °C, NaNO₃ might be being thermally decomposed in the present case. This was further confirmed by the TG/DTG analyses. As seen in Fig. S3A,† both as-prepared CoO_x/hSiO₂ and CoO_x/mSiO₂ samples showed initial weight losses from 100 to 300 °C, which can be ascribed to the evaporation of moisture on the catalyst surface, and the decomposition of cobalt precursor.⁵⁴ Interestingly, it is noted that there is a drastic weight loss from 300 to 800 °C on as-prepared CoO_x/mSiO₂ (Fig. S3A and B†), which can be attributed to the thermal decomposition of NaNO₃.⁵⁵ In spite of the rapid heating rates experienced by the droplet in aerosol process, the residence time of the droplet in the heated zone (5 s) was too short for complete decomposition of NaNO₃. And this was verified by the XRD analysis (Fig. 1A), where diffraction reflections of crystalline NaNO₃ were clearly observed. Consequently, one may deduce that only a certain amount of NaNO₃ was being thermally decomposed in aerosol process.

The decomposition products from NaNO₃, such as O₂, may further catalyze the oxidation of Co²⁺ yielding Co³⁺, accompanying the formation of spinel-structured Co₃O₄. Although oxidizing gases may be produced from the inherent decomposition of cobalt nitrate, the predominant existence of Co²⁺ species on both CoO_x/hSiO₂ and CoO_x/HMSP samples, as revealed by XPS studies, suggests that the impact of these low concentration gases on the nature of supported cobalt species is negligible. The above results clearly demonstrate that the NaCl and NaNO₃ salts, which were in situ formed during the acidification process of sodium silicate, played crucial roles on affecting the pore structure, oxidation state and surface reducibility of CoO_x/SiO₂ materials.

3.3 Toluene oxidation over various catalysts

In the blank experiment (without any catalyst), no conversion of toluene was detected below 500 °C under the conditions of toluene concentration = 1000 ppm and the GHSV = 20 000 h⁻¹, revealing that the direct thermal incineration of toluene was negligible under the adopted conditions. Fig. 4A displays the toluene conversion as a function of temperature over CoO_x/hSiO₂, CoO_x/mSiO₂, CoO_x/HMSP and Co₃O₄ catalysts. Obviously, the toluene conversion increased with the rise in reaction temperature and toluene was completely oxidized over all catalysts below 500 °C. The final products in the reaction were only CO₂ and H₂O, and no other by-products were detected.

Fig. 4 (A) Toluene conversion over (a) CoO_x/hSiO₂, (b) CoO_x/mSiO₂, (c) CoO_x/HMSP, and (d) Co₃O₄ (B) time-on-stream stability of toluene removal over CoO_x/mSiO₂ and Co₃O₄ catalysts at operation temperature of 250 °C. (C) Recycling tests for toluene removal over CoO_x/mSiO₂ at operation temperature of 250 °C. (Reaction condition: toluene concentration: 1000 ppm, toluene/O₂ molar ratio was 1/200 and SV: 20 000 h⁻¹).

The activities of the catalysts are evaluated by T₅₀ and T₉₀, corresponding to the temperature at 50% and 90% of toluene conversion, and decrease in the order of CoO_x/mSiO₂ (185 and 230 °C) > Co₃O₄ (230 and 245 °C) > CoO_x/hSiO₂ (260 and 290 °C) > CoO_x/HMSP (400 and 430 °C). For VOCs oxidation over supported cobalt catalysts, it is well-known that the catalytic performance is associated with a number of parameters of catalyst, such as the crystallinity of supported cobalt oxide, nature of the support, and reducibility of the cobalt species.^21,22,56 The support material having high surface area and large pore volume is reported to be favorable for dispersing active particles and improving the efficiency of the catalyst.^57,58 In this work, however, it is worth pointing out that the catalytic activity of CoO_x/HMSP is much inferior to those of CoO_x/hSiO₂ and CoO_x/mSiO₂, even though it consists of well-ordered mesostructure, and much higher surface area and pore volume (872 m² g⁻¹, 0.61 cm³ g⁻¹) than those of CoO_x/hSiO₂ and CoO_x/mSiO₂ samples (411 m² g⁻¹, 0.34 cm³ g⁻¹; 471 m² g⁻¹, 0.47 cm³ g⁻¹). In this sense, S_BET, V_pore and pore structure are minor factors for affecting the toluene oxidation in these supported cobalt catalysts.

The metal content is one of the crucial parameters that affect the activity of the supported catalyst. As shown in Table 1, the cobalt contents of the samples decrease in the order of CoO_x/HMSP > CoO_x/mSiO₂ > CoO_x/hSiO₂. However, the order of the cobalt content is not fully in line with the results of catalytic tests. It is therefore deduced that the cobalt content is not the key factor determining the activity of these catalysts. Alternatively, the UV-Vis spectra, XPS and H₂-TPR analyses suggest that the amount of spinel-structured Co₃O₄ with respect to Co³⁺ species decreases in the order of CoO_x/mSiO₂ > CoO_x/hSiO₂ > CoO_x/HMSP, which is perfectly consistent with the order of catalytic tests. According to the literature, it is known that the predominant formation of Co₃O₄ oxide phase with facile reducibility of Co³⁺ species on the support is advantageous for efficient VOCs oxidation,^16,17,24 which can well elucidate the highest catalytic activity of CoO_x/mSiO₂ catalyst. For CoO_x/hSiO₂ sample, its open-hollow structure with nanoporous shell might allow the toluene molecules access to the active cobalt sites more easily; however, its relatively weaker surface reducibility resulted in a higher reaction temperature.

On the opposite, the poorest performance of CoO_x/HMSP should be related to its highest amount of hardly reducible Co²⁺ and Co²⁺–silicate-like species, which are considered inactive in the studied reaction. Considering the results of structure–performance relationship, one can conclude that the presence of Co₃O₄ oxide phase with high amount of easily reducible Co³⁺ species is favorable for high catalytic activity on toluene oxidation. Besides, it is also found that the catalytic activity over CoO_x/mSiO₂ is higher than unsupported Co₃O₄ catalyst. Thus it is reasonable to suggest that mSiO₂ support shows a significant promoting effect for the activity of cobalt oxide catalysts, and this could be related with the higher dispersion and enhanced reducibility of cobalt oxide species on mSiO₂.^59,60

3.4 Stability of CoO_x/mSiO₂ catalyst

Fig. 4B depicts the time-dependent catalytic performance of CoO_x/mSiO₂ under conditions of 1000 ppm toluene at a temperature of T₁₀₀ (250 °C). Since unsupported Co₃O₄ showed almost complete removal of toluene at 250 °C (Fig. 4A), its durability was also investigated for comparison. Initially, the conversion of toluene over unsupported Co₃O₄ slightly decreased from 98.6% to 95.1% and can be well sustained during the following 16 h. Then it quickly decreased to 62% within 2 h and then completely deactivated after running for 29 h on stream. Conversely, the toluene conversion over CoO_x/mSiO₂ catalyst almost kept constant at 94% after running for 33 h, revealing its excellent tolerance and catalytic stability. Fig. 4C shows the reuse performance of CoO_x/mSiO₂ during 5 cycles of operation. It can be seen that the toluene conversion was very stable and exhibited within 1% attrition during the 5 cycles of operation. On the basis of the obtained results, one can conclude that the synthesized CoO_x/mSiO₂ possessed good stability and durability.

The XRD and nitrogen physisorption analyses of the used Co₃O₄ and CoO_x/mSiO₂ samples were tested and shown in Fig. S4 and Table S2.† The XRD patterns show that the crystallinity of used Co₃O₄ sample diminishes considerably as compared with that of fresh one. It is known that water is produced during VOCs oxidation and always has a negative effect on the catalyst structure.^61,62 In this work, the deactivation of unsupported Co₃O₄ catalyst could be associated with the collapse of active Co₃O₄ component by the presence of moisture. Conversely, there is no significant difference observed in the XRD and surface area results between fresh and used CoO_x/mSiO₂ samples. Moreover, the Co 2p region XPS spectra (Fig. S5†) also reveal that after the catalytic test in toluene oxidation, the signal intensities of used CoO_x/mSiO₂ only weakened slightly with respect to fresh sample, and the binding energies were nearly unchanged, indicating that active cobalt oxides can be well-stabilized and effectively preserved on mSiO₂ support during consecutive reaction. This high stability of the active cobalt oxide particles together with the mesostructure preservation are the key factors for the excellent activity of CoO_x/mSiO₂ catalyst.

4. Conclusions

By adopting a rapid and single-step salt-assisted aerosol method, hollow and mesoporous CoO_x/SiO₂ spheres were synthesized without using extra pore templates. The XRD, nitrogen physisorption, ICP-MS, UV-Vis, XPS, and H₂-TPR results clearly showed that the sodium salts, which were in situ formed during the acidification of sodium silicate solution, had significant impacts on the crystallinity, cobalt content, textural structure, oxidation state and surface reducibility of CoO_x/SiO₂ catalysts. The catalytic behaviors of CoO_x/SiO₂ catalysts were further investigated in catalytic oxidation of toluene at temperature range of 100–500 °C. It was demonstrated that mesoporous CoO_x/mSiO₂ performed better than the hollow CoO_x/hSiO₂, probably due to a combination of several factors of predominant existence of Co₃O₄ active phase, high surface Co³⁺ content, and easy reducibility of Co³⁺ at low temperature. This facile method could be extendable to the preparation of various hollow and mesoporous silica supported composites for a wide range of environmental applications.

Acknowledgements

The authors gratefully acknowledge the financial support from the Ministry of Science and Technology of Taiwan, R.O.C. through grant no.: NSC 102-2221-E-009-009-MY3.

References

1. H. Huang, Y. Xu, Q. Feng and D. Y. C. Leung, Catal. Sci. Technol., 2015, 5, 2649–2669.
2. J. W. Li, K. L. Pan, S. J. Yu, S. Y. Yan and M. B. Chang, J. Environ. Sci., 2014, 26, 2546–2553.
3. H. C. Wang, H. S. Liang and M. B. Chang, J. Hazard. Mater., 2011, 186, 1781–1787.
4. L. F. Liotta, H. Wu, G. Pantaleo and A. M. Venezia, Catal. Sci. Technol., 2013, 3, 3085–3102.
5. T. Garcia, S. Agouram, J. F. Sánchez-Royo, R. Murillo, A. M. Mastral, A. Aranda, I. Vázquez, A. Dejoz and B. Solsona, Appl. Catal., A, 2010, 386, 16–27.
6. M. F. M. Zwinkels, S. G. Järås, P. G. Menon and T. A. Griffin, Catal. Rev., 1993, 35, 319–358.
7. A. Y. Khodakov, J. Lynch, D. Bazin, B. Rebours, N. Zanier, B. Moisson and P. Chaumette, J. Catal., 1997, 168, 16–25.
8. L. F. Liotta, G. Di Carlo, G. Pantaleo, A. M. Venezia and G. Deganello, Appl. Catal., B, 2006, 66, 217–227.
9. D. L. Trimm, Catal. Today, 1995, 26, 231–238.
10. Z. Zhao, M. M. Yung and U. S. Ozkan, Catal. Commun., 2008, 9, 1465–1471.
11. M. M. Yung, Z. Zhao, M. P. Woods and U. S. Ozkan, J. Mol. Catal. A: Chem., 2008, 279, 1–9.
12. S.-H. Wu, C.-Y. Mou and H.-P. Lin, Chem. Soc. Rev., 2013, 42, 3862–3875.
13. C.-C. Li, U.-T. Wu and H.-P. Lin, J. Mater. Chem. A, 2014, 2, 8252–8257.
14. C.-C. Li, Y.-W. Chen, R.-J. Lin, C.-C. Chang, K.-H. Chen, H.-P. Lin and L.-C. Chen, Chem. Commun., 2011, 47, 9414–9416.
15. C.-K. Chen, Y.-W. Chen, C.-H. Lin, H.-P. Lin and C.-F. Lee, Chem. Commun., 2009, 46, 282–284.
16. Á. Szegedi, M. Popova and C. Minchev, J. Mater. Sci., 2009, 44, 6710–6716.
17. Z. Mu, J. J. Li, M. H. Duan, Z. P. Hao and S. Z. Qiao, Catal. Commun., 2008, 9, 1874–1877.
18. S. Zuo, F. Liu, J. Tong and C. Qi, Appl. Catal., A, 2013, 467, 1–6.
19. Z. Mu, J. J. Li, H. Tian, Z. P. Hao and S. Z. Qiao, Mater. Res. Bull., 2008, 43, 2599–2606.
20. T. Tsoncheva, L. Ivanova, J. Rosenholm and M. Linden, Appl. Catal., B, 2009, 89, 365–374.
21. H. Wu, G. Pantaleo, G. D. Carlo, S. Guo, G. Marcì, P. Concepción, A. M. Venezia and L. F. Liotta, Catal. Sci. Technol., 2015, 5, 1888–1901.
22. Z. Zhu, G. Lu, Z. Zhang, Y. Guo, Y. Guo and Y. Wang, ACS Catal., 2013, 3, 1154–1164.
23. J.-S. Girardon, E. Quinet, A. Griboval-Constant, P. A. Chernavskii, L. Gengembre and A. Y. Khodakov, J. Catal., 2007, 248, 143–157.
24. M. Popova, A. Ristić, V. Mavrodinova, D. Maučec, L. Mindizova and N. N. Tušar, Catal. Lett., 2014, 144, 1096–1100.
25. Z. Zhang, Y. Liang, Q. Ren, H. Liu and Y. Chen, Chin. J. Catal., 2011, 32, 250–257.
26. W.-N. Wang, J. Park and P. Biswas, Catal. Sci. Technol., 2011, 1, 593–600.
27. L. Liu, C. Zhao, H. Zhao, D. Pitts and Y. Li, Chem. Commun., 2013, 49, 3664–3666.
28. S. H. Choi, Y. J. Hong and Y. C. Kang, Nanoscale, 2013, 5, 7867–7871.
29. Y. J. Hong and Y. C. Kang, Nanoscale, 2014, 7, 701–707.
30. C. Zhao, L. Liu, H. Zhao, A. Krall, Z. Wen, J. Chen, P. Hurley, J. Jiang and Y. Li, Nanoscale, 2013, 6, 882–888.
31. N. E. Motl, A. K. P. Mann and S. E. Skrabalak, J. Mater. Chem. A, 2013, 1, 5193–5202.
32. C. Boissiere, D. Grosso, A. Chaumonnot, L. Nicole and C. Sanchez, Adv. Mater., 2011, 23, 599–623.
33. J. H. Bang and K. S. Suslick, Adv. Mater., 2010, 22, 1039–1059.
34. D. P. Debecker, M. Stoyanova, U. Rodemerck, F. Colbeau-Justin, C. Boissère, A. Chaumonnot, A. Bonduelle and C. Sanchez, Appl. Catal., A, 2014, 470, 458–466.
35. C. Gérardin, J. Reboul, M. Bonne and B. Lebeau, Chem. Soc. Rev., 2013, 42, 4217–4255.
36. L.-Y. Lin, J.-T. Kuo and H. Bai, J. Hazard. Mater., 2011, 192, 255–262.
37. L.-Y. Lin and H. Bai, Environ. Sci. Technol., 2013, 47, 4636–4643.
38. C. Y. Wang and H. Bai, Catal. Today, 2011, 174, 70–78.
39. C. Wang and H. Bai, Ind. Eng. Chem. Res., 2011, 50, 3842–3848.
40. J. Schlomach and M. Kind, J. Colloid Interface Sci., 2004, 277, 316–326.
41. H. Isobe, S. Utsumi, K. Yamamoto, H. Kanoh and K. Kaneko, Langmuir, 2005, 21, 8042–8047.
42. A. Prabhu, L. Kumaresan, M. Palanichamy and V. Murugesan, Appl. Catal., A, 2010, 374, 11–17.
43. C. Y. Jung, J. S. Kim, T. S. Chang, S. T. Kim, H. J. Lim and S. M. Koo, Langmuir, 2010, 26, 5456–5461.
44. M. Karthik, A. Faik, S. Doppiu, V. Roddatis and B. D'Aguanno, Carbon, 2015, 87, 434–443.
45. M. Karthik, E. Redondo, E. Goikolea, V. Roddatis, S. Doppiu and R. Mysyk, J. Phys. Chem. C, 2014, 118, 27715–27720.
46. M. Karthik and H. Bai, Appl. Catal., B, 2014, 144, 809–815.
47. T. Tsoncheva, G. Issa, T. Blasco, M. Dimitrov, M. Popova, S. Hernández, D. Kovacheva, G. Atanasova and J. M. L. Nieto, Appl. Catal., A, 2013, 453, 1–12.
48. S. H. Kim, B. Y. H. Liu and M. R. Zachariah, Chem. Mater., 2002, 14, 2889–2899.
49. B. Solsona, T. E. Davies, T. Garcia, I. Vázquez, A. Dejoz and S. H. Taylor, Appl. Catal., B, 2008, 84, 176–184.
50. Q. Wang, Y. Peng, J. Fu, G. Z. Kyzas, S. M. R. Billah and S. An, Appl. Catal., B, 2015, 168–169, 42–50.
51. Q. Tang, Q. Zhang, P. Wang, Y. Wang and H. Wan, Chem. Mater., 2004, 16, 1967–1976.
52. C. Wang, S. Lim, G. Du, C. Z. Loebicki, N. Li, S. Derrouiche and G. L. Haller, J. Phys. Chem. C, 2009, 113, 14863–14871.
53. X. Xie, Y. Li, Z.-Q. Liu, M. Haruta and W. Shen, Nature, 2009, 458, 746–749.
54. M. T. Makhlouf, B. M. Abu-Zied and T. H. Mansoure, Met. Mater. Int., 2013, 19, 489–495.
55. J. H. Peredes, H. E. Ponce, M. P. Beltran, M. E. Alvarez Ramos and A. D. Moller, Microsc. Microanal., 2007, 13, 740–741.
56. G. Bai, H. Dai, J. Deng, Y. Liu, F. Wang, Z. Zhao, W. Qiu and C. T. Au, Appl. Catal., A, 2013, 450, 42–49.
57. C. He, X. Zhang, S. Gao, J. Chen and Z. Hao, J. Ind. Eng. Chem., 2012, 18, 1598–1605.
58. C. He, Q. Li, P. Li, Y. Wang, X. Zhang, J. Cheng and Z. Hao, Chem. Eng. J., 2010, 162, 901–909.
59. J. Deng, L. Zhang, H. Dai and C.-T. Au, Appl. Catal., A, 2009, 352, 43–49.
60. Z. Qu, Y. Bu, Y. Qin, Y. Wang and Q. Fu, Appl. Catal., B, 2013, 132–133, 353–362.
61. D. R. Merrill and C. C. Scalione, J. Am. Chem. Soc., 1921, 43, 1982–2002.
62. C. Hu, Q. Zhu, Z. Jiang, L. Chen and R. Wu, Chem. Eng. J., 2009, 152, 583–590.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra27488e

---