Templated synthesis of crystalline mesoporous CeO₂ with organosilane-containing polymers: balancing porosity, crystallinity and catalytic activity

Crystalline mesoporous oxides with interesting redox and electronic properties have been broadly used in various applications in catalysis [1–5] and energy storage [6–9], given the large surface area and accessible pores in mesoscale (2–50 nm). In many cases, the crystallinity and crystalline phases of those mesoporous oxides are critical to their applications [10–14]. Taking mesoporous TiO₂ as an example, TiO₂ with the mixed phases of anatase and rutile can promote the charge separation at the two crystalline phases upon photoexcitation that enhances the overall photocatalytic performance [15, 16]. Synthesis of crystalline mesoporous oxides usually relies on high-temperature annealing to crystallize the framework of mesoporous oxides [17–21]. The key challenge of such an annealing process is to balance the porosity and crystallinity because the overgrowth of crystalline domains can cause the collapse of frameworks and elimination of the porosity [3, 22–25]. A common solution is the use of hard templates such as mesoporous silica and carbon [26–31]. Those hard templates, that are mesoporous to start with, can be filled with the precursor to grow the targeted oxides and subsequently removed after the formation of pores [21, 32–36]. Although the hard-templating approach provides a versatile tool applicable to a number of oxides, it often suffers from low efficiency and high cost as pointed out in literature [17].

Numerous efforts have been devoted to synthesizing crystalline mesoporous oxides through soft-templating method where the metal precursor and organic templates (e.g. polymers and surfactants) can self-assemble into ordered liquid crystalline forms in the sol–gel conversion [37–42]. Although soft organic templates are not thermally stable enough to confine the crystallization of oxides, previous studies suggest that conversion of those organic templates to inorganics, e.g. carbon [39, 43], can resolve the stability challenges to grow mesoporous oxides at high temperature. Soft-templating synthesis is advantageous because the pore size and the wall thickness of porous oxides are readily controllable, as compared to the hard-templating method. On the other hand, our group has developed the use of hybrid polymer templates to overcome the poor thermal stability of organic polymers. The method is simply based on the design of organosilane-containing block copolymers (BCPs), poly(ethylene oxide)-block-poly[3-(trimethoxysilyl)propyl methacrylate] (PEO-b-PTMSPMA) [44–50]. The PTMSPMA block consists of trimethoxysilyl moieties that can hydrolyze to form polysilsesquioxane and further thermally convert to inorganic silica-like structures [51, 52]. Very similar to soft-templating synthesis, the BCP PEO-b-PTMSPMA can assemble with metal precursors along with the sol–gel conversion during evaporation-induced self-assembly (EISA) [50]. At elevated temperature, the PEO block will decompose and the residual silica from the PTMSPMA block serves as 'hard' templates to grow crystalline oxides. Those organosilane-containing polymer templates combine the advantages of soft- and hard-templating syntheses to tune the porosity within crystalline oxide frameworks. Without the use of any inert gas atmosphere, our templates allow the thermal engineering to improve the crystallinity of porous frameworks up to 1000 °C in air. We have demonstrated the use of hybrid polymer templates to grow a few crystalline mesoporous oxides, e.g. TiO₂ [45, 50] and titanates [44]. While the interaction of Ti⁴⁺ ions and PEO has been well-understood, the synthetic strategy has been largely limited to Ti-containing crystalline oxides.

We herein report the facile synthesis of crystalline mesoporous ceria (mCeO₂) using hybrid polymer templates of PEO₁₁₄-b-PTMSPMA₂₂₄. Ceria has been a fundamentally important oxide used in a broad range of catalysis mostly as a support for metal catalysts [53–58]. The synthesis of mesoporous ceria, however, has been limited to the hard-templating approach only [59–62]. In the current synthesis, the delicate interaction of PEO and Ce³⁺ ions can be controlled by solvent composition of water and ethanol. By weakening interaction of PEO and Ce³⁺ ions, we show the co-self-assembly of hybrid polymer templates and Ce³⁺ ions during EISA. Those hybrid polymer templates can stabilize the porous frameworks of mCeO₂ while crystallizing at 900 °C. A similar synthetic strategy can be extended to mesoporous Co₃O₄. We further demonstrate crystallinity-dependent catalytic activity of mCeO₂ using electrochemical oxidation of sulfite.

2.1. Materials

2.2.  Synthesis of hybrid polymer templates and mCeO₂

2.3. Characterization

2.4. Electrochemical measurements

Synthesis of mCeO₂ was templated by polymer micelles of PEO_114- b-PTMSPMA₂₂₄ according to our previous reports [50, 51]. Polymer micelles of PEO₁₁₄-b-PTMSPMA₂₂₄ were first prepared by self-assembly in the water/ethanol mixture. The PTMSPMA core could be further hydrolyzed in the presence of trimethylamine to form silica-like NPs. The diameter of polymer micelles is 22.3 ± 3.2 nm (scheme 1). The solution of polymer micelles was further dialyzed against ethanol to remove residual triethylamine and water. The preparation of mCeO₂ was carried out in ethanol/water mixture through EISA where the coordination of PEO and Ce³⁺ ions led to the formation of gel. The ethanol/water mixture was used to tune the interaction strength between PEO and Ce³⁺ ions. In pure ethanol solution, PEO and Ce³⁺ ions interacted strongly where the addition of Ce(NO₃)₃ to the micellar solution resulted in precipitation of micelles. In pure water, PEO and Ce³⁺ ions interacted weakly. The mixing of Ce(NO₃)₃ and the micellar solution did not result in any apparent change in solution transmittance. We varied the concentration of water in the mixture, as a means to increase the solvation of Ce³⁺ ions and decrease the interaction of PEO and solvated Ce³⁺ ions (see below), so that a homogeneous solution of Ce(NO₃)₃ and the micellar solution was formed initially. The slow removal of solvents would allow the self-assembly of polymer micelles and the Ce salt. After drying at 40 °C overnight, the sol of Ce(NO₃)₃ and the micellar solution formed a homogeneous lightly-yellow gel. The gel was collected and calcined at elevated temperatures to crystallize CeO₂. After thermal annealing, the residual silica was etched by hot NaOH to expose the pores and yield the yellow powder as CeO₂ (figure S1). We denoted our mesoporous CeO₂ as mCeO2-X where X represents the calcination temperature.

Zoom In Zoom Out Reset image size

Using mCeO₂-700 as an example, 90 mg of Ce(NO₃)₃ (50 mM in ethanol) was mixed with 8 ml of polymer micelles (15 mg ml⁻¹ in ethanol). After adding 2 ml of water, the solution became transparent. The homogeneous solution was kept stirring at room temperature for 1 h and then transferred to a petri dish to remove solvents. The dried gel was further annealed at 100 °C for 24 h and ramped to 700 °C at a heating rate of 2 °C min⁻¹ under air. After calcination for 4 h, the samples were washed with 2 M NaOH (55 °C) to remove residual silica. The crystalline porous frameworks of mCeO₂-700 were first revealed by electron microscopy (figure 1). Figure 1(a) shows a low-resolution TEM image of mCeO₂-700. Highly porous and ordered mesoporous structures were seen throughout the large piece of crystals. The pores could be seen under zoom-in TEM images in figure 1(b) and the size of pore is in the range of 10–20 nm with the typical measurement results in the image. SEM confirms the porous structures on the surface of bulky pieces (figures 1(c) and (d)). The pore size analyzed from has a narrow distribution with an average pore size of 13.0 ± 1.6 nm, in consistence with TEM observation. The size of pores was obviously smaller than the size of polymer micelles. This is due to the shrinkage induced by high temperature as reported previously [50, 63–66]. The highly crystalline framework was confirmed using HR-TEM. Figures 1(e) and (f) displays the porous structure and clear lattice fringes of CeO₂. The d-spacings of 0.31 nm and 0.27 nm were assigned to (111) and (200) lattice planes of CeO₂, respectively. The crystalline nature of the frameworks was also confirmed using XRD (figure 1(i)). Only cubic fluorite type CeO₂-phase was seen without other impurities (JCPDS no. 34–0394). The crystalline grain size estimated from Scherrer equation is 5.6 nm. The yellow mCeO₂-700 has a light-yellow color with its optical bandgap of 3.1 eV as measured from diffusion reflectance UV–vis spectroscopy [67] (figure S2).

Zoom In Zoom Out Reset image size

The porosity was confirmed using nitrogen sorption experiments by BET analysis. The sorption isotherm shows a characteristic type-IV hysteresis loop. The specific surface area of mCeO₂-700 is 84.4 m² g⁻¹ (figure 1(g)). The pore size analyzed from BJH method is 14.0 nm, close to that measured from electron microscopy. As a control, we prepared bulk CeO₂ in the absence of polymer templates using a similar procedure. Bulk CeO₂ does not show significant nitrogen sorption with a low surface area of 2.6 m² g⁻¹ (figure S3).

The key parameter to control the synthesis of mCeO₂ is the solvent ratio to tune the interaction of Ce³⁺ ions and polymer micelles. Unlike previous synthesis of TiO₂ [50] and WO₃ [25], Ce³⁺ ions show strong binding with PEO in ethanol. As controls, we examined the mixing of Ce(NO₃)₃ and polymer micelles in water/ethanol mixture with different ratios. The solution turbidity as measured by the transmission of the solution had an obvious dependence on water concentration. In the absence of water, the solution was completely turbid with its transmission <3% at 600 nm measured from UV–vis (figures S4(a) and (b)). The addition of water resulted in the decrease of turbidity. At ∼5 vol% of water, the solution became completely transparent. While, the solution kept turbid diluted by ethanol as control (figures S4(c) and (d)). As for EISA, we dried the sol with different water concentrations. Homogeneous gels were observed with a water concentration in the range of 10 vol% to 20 vol%. In additional, the use of PEO₁₁₄-b-PTMSPMA₂₂₄ is of critical importance to preserve the pores of mCeO₂ at high temperature under air [44, 50]. Other soft templates like poly(ethylene oxide)-block-polystyrene do not provide similar thermal stability under air [68–70]. We can further extend our synthetic method to synthesize other transition metal oxides, like Co₃O₄ (figure S5), challenging to synthesize using the traditional EISA [71].

As enabled by the thermal stability of polymer templates, we further examined the structural and crystalline evolution of mCeO₂ in the calcination temperature range of 400 °C–900 °C for 4 h (see figure S6). Figure 2 shows the typical SEM and BET results of mCeO₂ obtained at 400 °C (figures 2(a) and (b)), 600 °C (figures 2(c) and (d)), and 900 °C (figures 2(e) and (f)). All mCeO₂ shows remarkable thermal stability. The porous frameworks retained their structural integrity even after calcined at 900 °C. Below 800 °C, the pores were well-preserved and the average size of pores was in the range of 10–15 nm as estimated from SEM (table 1). Disordered pores with a larger polydispersity were seen for mCeO₂-900. This was likely because of the thermal shrinkage of the framework at a higher calcination temperature. Nevertheless, all samples show typical type-IV isotherm with a characteristic hysteresis loop (figure 2). The pore size of mCeO₂ from BET is also in the range of 10–14 nm when the calcination temperature is below 800 °C. There is a slightly decrease of pore size for mCeO₂-900 to 6 nm along with the decrease of surface area. This is likely due to the overgrowth of CeO₂ crystallites at high temperatures [72–74].

Zoom In Zoom Out Reset image size

The crystallinity of mCeO₂ was characterized using XRD. Figure 3(a) show the XRD patterns of mCeO₂ obtained at different calcination temperature. The diffraction peaks at 28.6°, 33.1°, 47.5°, 56.3°, 59.1 , 69.4°,76.7° and 79.1° are assigned to the (111), (200), (220), (311), (222), (400), (331) and (420) facets of cubic fluorite structured CeO₂, respectively. With the increase of calcination temperature, there is no new peak, indicating the stable phase of cubic fluorite-type CeO₂ without phase transition during thermal annealing. The elevation of the calcination temperatures had a significant impact on the XRD peak broadness. At lower temperature, broad diffraction peaks were seen and no clear peaks were observed for high-indexes facets. While, the peaks became shaper above 600 °C. This is indicative of improved crystallinity of mCeO₂ frameworks. Qualitatively, the crystalline grain size calculated from Scherrer equation shows an obvious increase from 4.5 nm for mCeO₂-400 to 12.2 nm for mCeO₂-900 (figure 3(b)). The diffraction peaks at 69.4°, 76.7° and 79.1° became more pronounced when the calcination temperature above 800 °C, as crystalline domain grew larger. This is similar to our previous observation on TiO₂. Polymer templates of PEO₁₁₄-b-PTMSPMA₂₂₄ were thermally converted to inorganic silica nanoparticles that are mechanically strong to prevent the overgrowth of CeO₂ crystallites on the frameworks [75]. While the temperature raised beyond the glass transition of silica, the sintering of silica templates would result in the collapse of pores [76].

Zoom In Zoom Out Reset image size

The change in crystallinity of mCeO₂ at different calcination temperature was also confirmed by Raman scattering. Polycrystalline ceria shows a very pronounced Raman scattering peak at 463 cm⁻¹ assigning to the F_2g band. Previous studies suggested that the peak shift and linewidth of the F_2g band were dependent on the crystallinity and crystalline grain size [77, 78]. Increasing calcination temperature sharps the F_2g band of ceria [79]. The peak half width at 463 cm⁻¹ is plotted as a function of temperature in figure 3(d) where a nearly linear correlation was found. Those results suggest the improved crystallinity of mCeO₂ at higher calcination temperature. Note that, there is a slightly blue shift of the F_2g band to 464 cm⁻¹, likely due to the formation of oxygen vacancies as a result of oxygen release [77].

Crystalline engineering can also be carried out through increasing thermal annealing time. At 700 °C, we systematically examined crystallization of mCeO₂ by increasing the annealing time from 1 h to 48 h. Despite of similar diffraction patterns (figures S7(a) and (b)), the improved crystallinity of ceria was clear as the peaks became sharp. The crystal grain size shows a very small increase from 5.0 nm (mCeO₂-700-1 h) to 5.6 nm for mCeO₂-700-4 h. Interestingly, further increasing the calcination time had a limited impact on the crystal grain size. For mCeO₂-700-48 h, the crystal grain size remained to 6.1 nm (table 1). The peak half width of the F_2g band of ceria showed a continuous decrease along the thermal annealing (figures S7(c) and (d)). This is indicative of the thermal stability of polymer templates. While the thermal treatment enhances the crystallinity, the crystals are confined by the thermally stable and mechanically strong templates. The BET surface areas and total pore volume of mCeO₂-700 at different annealing times are shown in figure S8. All mCeO₂-700 samples exhibited typical mesoporous isotherms. The surface area was largely preserved at longer annealing time. mCeO₂-700-12 h exhibited a high surface area of 96 m² g⁻¹; while that of mCeO₂-700-48 h slightly decreased to 63 m² g⁻¹. Similar results were seen from SEM (figure S9).

We further evaluated the catalytic activity of mesoporous ceria for electrochemical oxidation of sulfite. Sulfite usually produced from flue gas desulfurization [80] is an excellent scavenger readily used in photocatalysis. We first used the LSV to examine the activity of mCeO₂ and bCeO₂ deposited on pyrolyzed graphite (PG). Figure 4(a) shows the typical LSV curves at a scan rate of 10 mV s⁻¹ in 0.5 M of phosphate buffer (pH ∼7) containing 1 M of Na₂SO₃ [81]. mCeO₂-700 is very active with a much higher anodic current density (j), as compared to the bare PG electrode and that with bCeO₂. At 1.0 V vs RHE, mCeO₂ has a j of 32 mA cm⁻², that is roughly 48 times higher than that of bCeO₂. The improved catalytic activity is attributed to the large surface area and porosity of mCeO₂. To confirm the activity arising from the oxidation of sulfite, LSVs at different concentrations of Na₂SO₃ were also collected (figure 4(b)). In comparison with the background scan at 0 M of Na₂SO₃, the anodic oxidation current increased with the concentration of Na₂SO₃. This is similar to other reports of n-type semiconductors showing similar catalytic oxidation of Na₂SO₃ [82].

Zoom In Zoom Out Reset image size

The activity of mCeO₂ had an obvious correlation with its crystallinity. When the calcination temperature is below 700 °C, improving crystallinity enhances the activity of mCeO₂. For mCeO₂-400, the j is 9 mA cm⁻² at 1 V and the j of mCeO₂-600 increased to 13.9 mA cm⁻² that is about 1.5 times compared to that of mCeO₂-400. While mCeO₂-700 has much better activity as compared to those samples calcined at low temperature, further increase of the calcination temperature would also decrease the activity of mCeO₂. For example, mCeO₂-900 has a j of 5.1 mA cm⁻² at 1 V, only 15% of mCeO₂-700 (figures 4(c) and (d)). While mCeO₂-900 would form larger crystallinity than mCeO₂-700, we attributed the loss of activity to the increase of grain sizes (figure 3(b)). The increase of crystalline grains would lead to the overgrowth of ceria that exposes less Ce sites on the surface. This also be confirmed from the potential to reach j of 15 mA cm⁻² (figure 4(e)). For mCeO₂-700, 0.96 V is needed to reach j of 15 mA cm⁻²; while, that of mCeO₂-400 increased to 1.04 V.

While crystallinity is critical for the activity of mCeO₂, the annealing time also shows an impact on the activity (figure 5(a)). mCeO₂-700-1 h that has a j of 6.1 mA cm⁻² at 1 V was less active compared to mCeO₂-400-4 h. The activity of mCeO₂-700-12 h was among the highest with a j of 42.3 mA cm⁻² at 1 V, approximately 7 times more active as compared to mCeO₂-700-1 h and 62 times more active as compared to bCeO₂ (figure 5(b)). Further increasing the calcination time would lead to the decrease of activity, similar to those at a higher temperature. It is presumably due to the decrease of surface area at a longer calcination time.

Zoom In Zoom Out Reset image size

EIS was used to study the charge transfer kinetics. Figure 6 shows the EIS results measured in saturated Na₂SO₃ solution at 1 V vs RHE. The equivalent circuit fits perfectly with the Nyquist plots for all mCeO₂ (figure 6(b)). Not surprisingly, mCeO₂-700 exhibits the lowest electron transport impedance with the charge transfer resistance (R_ct) of 35.7 Ω (figure 6(c)). The trend in R_ct was aligned with the catalytic activity of mCeO₂. Below 700 °C, high-temperature annealing improves the charge transfer rate constant, which is inversely proportional to R_ct. The heterogeneous apparent electron transfer rate constant (k_app) characterizes the intrinsic activity of electrochemical active species, and k_app was calculated from [83, 84]:

$$\begin{equation*}{k_{{\text{app}}}} = \frac{{RT}}{{{n^2}{{F}^{\,2}}A{R_{{\text{ct}}}}C}}\end{equation*}$$

Zoom In Zoom Out Reset image size

where the R is the gas constant (8.314 J K⁻¹ mol⁻¹), T is the temperature (298 K), n is the number of electrons transferred in the reaction, which is n = 2 for electrooxidation of SO₃ ²⁻ to SO₄ ²⁻. F is Faraday's constant, R_ct is the charge transfer resistance in Ω, A is the area of the working electrode, and C is the concentration of Na₂SO₃. The results are summarized in figure 6(c). mCeO₂-700 and mCeO₂-800 have a close k_app of 44.4 $\times$10⁻⁷ cm s⁻¹ and 42.8 $\times$ 10⁻⁷ cm s⁻¹, respectively, that is close to 2 times as compared to mCeO₂ calcined at a lower temperature. Increasing the crystallinity while retaining their porosity is of critical importance. mCeO₂-700-12 h shows a k_app of 101.1$\times$10⁻⁷ cm s⁻¹ suggesting that the electron transfer rate at the surface of mCeO₂ is about 5-times faster than for less crystalline samples like mCeO₂-400. In addition, the porosity of mCeO₂ contributes significantly to the activity. As compared to bCeO₂, mCeO₂-700-12 h shows about 84 times faster charge transfer rate. Other than the improvement of crystalline, the surface oxidation state of Ce may contribute to the overall catalytic performance. Although it is close to the commercial one, mCeO₂-700-1 h and mCeO₂-700-12 h still showed Ce³⁺ of 28 and 32 at%, respectively, as measured from XPS (see figure S10). The increase of Ce³⁺ ions on the surface is indicative of surface oxygen vacancy that usually enhances the binding of substrates. Therefore, the balance of crystallinity and porosity (surface area) of mCeO₂ catalysts contributes to the improved electrochemical catalytic activity.

The results above demonstrate a facile and useful method to synthesize ordered mesoporous ceria with high crystallinity and thermal stability. The use of organosilane-containing polymer micelles could provide thermally stable and mechanically strong templates to prevent the collapse of mesoporous frameworks during calcination. We showed the simple EISA process by controlling the interaction between templates and metal precursors. At 700 °C, mCeO₂-700 has a specific surface area of 84 m² g⁻¹ with a pore size of 14 nm. mCeO₂ was active for electrochemical oxidation of sulfite. With porosity, mCeO₂-700 showed a higher anodic current density of 32 mA cm⁻² at 1 V vs RHE, roughly 48 times higher than that of bCeO₂. Using EIS, we measured the intrinsic activity of mCeO₂ through apparent electron transfer rate constant. mCeO₂-700-12 h with a perfect balance of crystallinity and porosity had an electron transfer rate of 10⁻⁵ cm s⁻¹, that is about 84 times faster compared to bCeO₂ and 5 times faster compared to mCeO₂ with a lower crystallinity.

J H is grateful for the financial support from National Science Foundation (CEBT-1705566). The HR-TEM studies were performed using the facilities in the UConn Thermo Fisher Scientific Center for Advanced Microscopy and Materials Analysis (CAMMA). Low-magnification TEM and SEM images were obtained at the Biosciences Electron Microscopy Facility at the University of Connecticut. This work was also partially supported by the Green Emulsions Micelles and Surfactants (GEMS) Center of the University of Connecticut.