Highly porous Ce–W–TiO₂ free-standing electrospun catalytic membranes for efficient de-NO_xvia ammonia selective catalytic reduction

Abstract

Highly porous Ce–W–TiO₂ free-standing nanofibrous membranes (FSM) are fabricated via electrospinning techniques to serve as NO_x-SCR catalysts. The precursor of the ceramic nanofibers (sol–gel solution) is co-electrospun with poly(vinyl alcohol) (PVA) water solution. PVA integration into FSM has been proven to avoid excessive bending of the nanofibers and to prevent mechanical failure of the final ceramic nanofibrous structure. This is demonstrated to be associated with higher thermal stability of PVA compared with that of other organic additives. 3D tomography reconstruction indicates a resulting ceramic membrane with remarkable open and interconnected porosity of ca. 96%. Catalytic characterization, performed at the best working conditions (in absence of H₂O and SO₂), indicates that amorphous FSM is the best performing catalytic membrane. Superior catalytic performances of the developed FSM over those of other nanofibers and nanoparticle catalysts are proven because of superior surface, morphological, and structural features. Long-term stability (120 h) and reproducibility (over 5 cycles) of FSM are also demonstrated.

---

Environmental significance

Nitrogen oxides (NO_x) are recognized as major air pollutants responsible for serious damage to the environment and human health. The most efficient and cost-effective technology available to convert NO_x into harmless products (nitrogen and water) is the selective catalytic reduction (SCR) method. Nanometric metal oxides electrospun as membranes hold great potential for their use as enhanced catalysts due to their merits such as high surface-to-volume ratios, wide-open and interconnected porous networks, and high porosity. However, the preparation of free-standing electrospun ceramic membranes is critical because of their mechanical integrity. Herein, by integrating a co-electrospun polymeric component, we demonstrate highly porous WO₃–CeO₂/TiO₂ nanofibrous free-standing membranes, which exhibit superior de-NO_x catalytic performances compared with nanofibers at the same composition.

Introduction

Free-standing porous membranes are crucial for a wide range of practical applications that include biomedical engineering, water treatment, air purification, and catalysis.^1–5 Typically, free-standing membranes are fabricated from polymers to ensure inherent flexibility and mechanical stability.^6,7 However, when harsh conditions are used or operation at temperatures above 100–200 °C is needed, ceramic membranes offer a wider spectrum of options and possibilities.^6,8 For instance, materials are usually subjected to extreme environmental conditions during catalytic applications.^6,8 Ceramic porous membranes are typically prepared via colloidal processing that requires the dispersion of ceramic powder into a solvent with the aid of an adequate dispersant and the subsequent addition of a number of different additives to adjust the processability^9–12 and eventually impart porosity.^13,14

Besides the colloidal approach, in the recent years, electrospinning has emerged as a simple, inexpensive and versatile method for shaping materials into non-woven membranes composed of continuous nanofibers.^15–20 The resulting nanofibrous membranes possess significantly high porosity (e.g., 80–90% (ref. 21)) that is associated with an arrangement of the nanofibers into a wide-open and well-interconnected porous network.^21–25 As a class of one-dimensional (1D) nanomaterials with cross-sectional diameters ranging from tens to hundreds of nanometres, nanofibers further offer extremely high surface area-to-volume ratios. All these morphological features have been demonstrated to impart outstanding performances to the resulting nanofibrous ceramic membranes, especially in pressure-driven applications, such as for instance filters for the abatement of air pollutants. This is because the dimensions of the nanofibers match both the dimensions of particles to be captured and those of the molecular air flow parameters (e.g., λ, gas mean free path, for air at standard conditions is 68 nm).^24,26 In addition, the electrospinning approach offers the advantage of easy control over the chemical compositions, allowing the preparation of a vast number of materials by simply adjusting the chemistry of the starting spinning solution.²⁷

Despite many advantages, the preparation of free-standing thick ceramic electrospun membranes still remains critical because of the mechanical fragility of the nanofibers when flexed above their critical strain limits. Typically, titania nanofibers from sol–gel chemistry exhibit significant flexibility with decreased tensile strength and increased Young's modulus upon increasing calcination temperatures.²⁸ However, the transition from the as-spun material, containing both organic and inorganic precursors, to the final polycrystalline metal oxide is accompanied by shrinkage of the materials, which can lead to strains easily above 50% of the initial dimensions.²⁸ This might cause excessive bending of the single nanofibers, thus leading to fracture of the membrane into small fragments. A number of investigations have been conducted to prepare ceramic free-standing electrospun media.^3,4,16,27 For instance, Wang et al. successfully prepared self-standing γ-alumina fibrous membranes by careful optimization of the spinning parameters and of the calcination conditions.⁴ Furthermore, Shahreen et al. proposed combining the fragile catalytic nanofibers with more mechanically stable micron fibres of alumina into fibrous catalytic media for the removal of NO and CO. The active nanofibers and the supporting microfibers were separately electrospun, calcined, and then mixed to form a water-based slurry that was finally dried under vacuum into a mold to form a cylindrical filter.²⁴

Despite the promising performances obtained, the mentioned processing methods still suffer from limitations, the major one being the multi-step approach. A simple manufacturing method is highly desirable to reduce the production costs and time.

Herein, we address the issue by proposing a single-step approach based on the integration of an additional co-electrospun polymeric component into the nanofibrous structure, which can sustain catalytic nanofibers during the removal of the organics. Structural support is achieved by introducing a polymer with higher thermal stability compared to that of organic additives used in the electrospinning process. As a supporting polymer, we propose poly(vinyl alcohol) (PVA) that is co-electrospun together with precursor solutions of Ce–W–TiO₂ into nanofibrous xerogel membranes. Upon calcination, highly porous free-standing electrospun membranes (FSMs) are produced, which serve as catalytic media for the abatement of NO_xvia selective catalytic reduction (SCR). The composition of these ceramic nanofibers is fixed based on our recently reported results on the de-NO_x efficiency of ceria-based catalysts.²² To the best of our knowledge, this is the first report on the preparation of highly porous free-standing electrospun ceramic membranes using a supporting co-electrospun polymeric component to control the structural stability of ceramic nanofibers.

Experimental details

Materials

All materials were of reagent grade and used as received. Precursors sensitive to environmental humidity were stored in a desiccator. Titanium(IV) isopropoxide (97% purity, Sigma Aldrich, Denmark), tungsten(VI) ethoxide (5% w/v in ethanol, Alfa Aesar, Denmark), and cerium(III) nitrate hexahydrate (99.5% purity, Alfa Aesar, Denmark) were used as cationic precursors. Poly(methyl methacrylate) (PMMA; M_w ∼996 000, Sigma-Aldrich, Denmark) and glacial acetic acid (99.8%, Sigma Aldrich) were used as carrier polymer and stabilizer, respectively, for the hydrolysis of alkoxides. Ethanol (anhydrous, 99.9%, Sigma Aldrich, Denmark) and acetone (99.8% purity, VWR International, Denmark) were used as solvents. Poly(vinyl alcohol) (PVA, 87–89% hydrolyzed; M_w ∼85 000–124 000, Sigma Aldrich, Denmark) and distilled water were used to prepare the spinning solution of the supporting polymer.

Preparation of spinning solution

In an optimized procedure, the spinning solution for the ceramic nanofibers was prepared by first stabilizing titanium(IV) isopropoxide and tungsten(VI) ethoxide with acetic acid (e.g., molar ratio of acetic acid to alkoxides was kept at 15) under magnetic stirring at 20 °C in a sealed container purged with an inert gas (e.g., argon) for 15 min. A salt solution of cerium(III) nitrate hexa-hydrate in anhydrous ethanol at a concentration of 39% w/w was added dropwise into the stabilized mixture under mild stirring until a transparent and homogeneous solution was obtained. The solution viscosity was then adjusted by adding a required amount of PMMA solution (7.5% w/w). The overall cationic concentration was kept at 0.4225 mol L⁻¹, whereas the ratio of PMMA to cations was fixed at 0.56 g g⁻¹. The molar ratio of titanium/cerium/tungsten was fixed at 1.0/0.2/0.1. The PVA solution, at the concentration of 7.5% w/w, was prepared by dissolving PVA into distilled water. The two spinning solutions (i.e., the ceramic precursor and the polymer support solutions) were finally loaded into two different glass syringes equipped with 21-gauge stainless steel needles.

Electrospinning of the solution and heat treatment of the as-spun nanofibrous mats

The nanofibrous membranes were fabricated using RT Advanced (Linari Engineering, Italy). The two spinning solutions, loaded into two glass syringes, were connected to two separate syringe pumps and co-electrospun using the same spinning conditions of voltage (40 kV), tip-to-collector distance (12 cm), and solution-feeding rate (0.50 mL h⁻¹). The electrospinning process was conducted in air at a temperature of 20 °C with relative humidity controlled at 23%. The nanofibers were collected as a flexible green membrane on an aluminum foil wrapped on the grounded drum rotating at 100 rpm. The total volume of both the cationic and the polymer solutions was fixed at 10 mL for each fabricated membrane. The resulting green membranes were then thermally treated in air for 1.5 h at different temperatures (>500 °C) at a heating rate of 5 °C min⁻¹ to promote the removal of organic additives and favor the formation of desired crystallographic phases.

Material characterization

The thermal properties of PVA, PMMA and as-spun nanofibrous membranes were investigated (TGA, NETSZCH, STA 409, CD) in air flux from room temperature to 700 °C at a heating rate of 1 K min⁻¹. BET (Brunauer–Emmett–Teller; Autosorb 1-MP, Quantachrome Instruments, Boynton Beach, Fl, USA) analysis was conducted to measure the specific surface areas (SSA) of materials. The pore volume and pore diameter of the fibrous developed membranes were evaluated using the Barrett–Joyner–Halenda (BJH) method (AUTOSROB software, AS1WIN). X-ray diffraction (XRD, Bruker D8, Germany) patterns were recorded at room temperature (Cu Kα radiation) at a scanning rate of 0.01° s⁻¹ in the 2θ range from 10 to 90°. The morphology of nanofibers was analyzed using field emission scanning electron microscopy (FESEM, Supra, Carl Zeiss, Germany) and transmission electron microscopy (TEM, JEM3000F, Oxford Instruments, UK). The chemical composition of samples was determined by energy-dispersive spectroscopy (EDS) coupled with FESEM.

The 3D structure of the membrane was evaluated by FIB/SEM (Zeiss 1540 XB Crossbeam Scanning Electron Microscope, Carl Zeiss Microscopy GmbH, Germany) using a lateral E–T (Everhart–Thornley) detector and an In-lens detector. A gallium FIB slicing probe of 1 nA was used and the thickness of each slice was estimated to be 13 nm. Serial sectioning imaging was performed at 1 kV with a pixel size of 24 × 24 nm², yielding a voxel size in the 3D-data set of 13 × 24 × 24 nm³. The nanofibrous membrane was prepared for FIB/SEM microscopy by vacuum infiltration with epoxy resin (from Struers) to enable high-quality grinding and polishing of the sample. Segmentation of the 3D FIB/SEM image data was performed with the ImageJ program (NIH). The uneven illumination observed in the 3D dataset does not allow setting of a single threshold for the entire micrographs. Thus, the Sauvola algorithm^29,30 was used to perform local thresholds for the data. This algorithm divides the input image into square windows (n × n pixels) and sets thresholds for each window based on the mean and standard deviation of the pixel intensities. Visualization of the 3D reconstruction of the analyzed data was then performed with the Avizo program (FEI). Pore size distribution (PSD) was calculated according to the method described by Münch et al.:³¹ the segmented 3D volume is filled with spheres of a given radius. By reducing the radius incrementally, more volume will be filled. The PSD is then obtained by correlating the incrementally filled volumes with the corresponding radii. The fiber length distribution (FLD) was calculated based on the average path length method between two opposite vertices. However, from a thorough analysis of the 3D tomogram, some nodes are also observed. Here, two or more adjacent fibers cross each other, leading to unavoidable uncertainties in the FLD calculations.

SCR tests

Catalytic experiments were conducted at dry conditions and in the absence of SO₂ to characterize the intrinsic catalytic behaviour of the ceria-based catalysts. The setup for the SCR activity measurements is described in our previous studies.^21–23 Different catalytic membranes were cut using pressing mold to fit the quartz reactor tube size (inner diameter 15 mm). To attain the desired GHSV values, in the reactor, the total gas flow rate was varied (3, 6, 9, and 12 L h⁻¹). Inlet gas concentrations were fixed as follows: NO = 1000 ppm, NH₃ = 1000 ppm, O₂ = 8%, and CO₂ = 5%, balanced with argon gas. During the experiments, temperature was increased stepwise from 100 to 500 °C, whereas the NO concentration was continuously monitored by a mass spectrometer (Pfeiffer Vacuum, OmniStar™ GSD 301). The reaction system was maintained for 1 h at each reaction temperature to reach a steady state before performing the analysis of its catalytic performance.

Results and discussion

Free-standing nanofibrous membranes (hereafter, referred to as FSM) are fabricated by co-electrospinning the organic-precursor solutions of Ce–W–TiO₂ with an aqueous solution of PVA. The pure polymeric component (PVA) is herein selected to mechanically support the ceramic nanofibers during calcination and to avoid fragmentation of the final ceramic membrane. The desired effect is achieved because of the higher thermal stability of PVA than that of other organic additives used in the electrospinning process. Fig. 1a shows the thermogravimetric profiles together with the mass loss rate derivatives for PVA (grey dash) and nanofibers co-electrospun with PVA (FSM, grey line) and without PVA (NF, black line). PVA is a semi-crystalline biocompatible and biodegradable polymer largely used for its excellent film-forming, emulsifying, and adhesive characteristics.^32,33 Organic additives used in the process also include alkoxides of titanium and tungsten (their thermal removal is in the range of 200–250 °C) and poly(methyl methacrylate) (PMMA), used as a spinning agent, which typically decomposes at temperatures lower than 500 °C (via de-polymerization to monomers/monomer evaporation³⁴). The TGA/DTG profiles of the NF sample indicate complete removal of organic additives (alkoxides and PMMA) at approximately 300 °C, whereas nanofibers co-electrospun with PVA (FSM) show complete decomposition at a temperature slightly lower than 500 °C. The observed shift in the decomposition temperature is likely due to PVA, and its complete thermal removal is identified at around 500 °C (Fig. 1a). This experimental observation indicates that when the decomposition of organic additives is already occurring, PVA is not yet completely decomposed. Also, PVA likely confers flexibility to the nanofibrous structure, supporting the nanofibers during the thermal removal of organic spinning additives and avoiding their excessive bending. The subsequent complete removal of PVA has no effect on the mechanical integrity of the resulting organic-free catalytic nanofibers because of the inherent mechanical stability of titania nanofibers.²⁸ In the DTA profiles of NF and FSM (Fig. 1b), a light exothermic shoulder between 550 and 600 °C is observed, which is likely related to the formation of a crystallographic phase (e.g., anatase).

Fig. 1 (a) TG/DTG and (b) DTA profiles of PVA (grey dash), FSM (grey line), and NF (black line).

Based on these results, nanofibrous FSMs are calcined at temperatures higher than 500 °C to ensure the removal of any residuals. As indicated by the TGA profile, during the calcination process, green FSMs undergo considerable weight loss of around 75%, which clearly corresponds to an equally significant shrinkage of their volume. The latter can be empirically evaluated by measuring the membrane dimensions before and after calcination. Fig. 2a–d show the variation in lateral dimensions and thickness post-calcination at 500 °C. The ability of PVA co-electrospinning approach to provide the required structural stability is verified on a relatively large membrane, which represents the worst instance of internal mechanical stresses. A thermal conversion from green to pure ceramic nanofibrous FSM is indeed accompanied by contraction in volume of around 78%. It is worth noticing that despite significant shrinkage, the nanofibrous membrane retains its shape and above all its structural integrity. Moreover, a long range (hundreds of microns) elemental analysis (EDX, inset Fig. 2c) of the calcined nanofibers indicates a percentage molar ratio for Ti/Ce/W (1.0000/0.1822/0.0723), which is consistent with the molar ratio in the initial precursor solution, suggesting adequate dispersion of the cations both in the initial spinning solution and in the electrospun membranes.

Fig. 2 (a) Digital photo of the produced FSM and (b) its SEM cross-sectional view; (c) digital photo of FSM calcined at 500 °C and (d) its SEM cross-sectional views. The inset in (c) represents EDX analysis for the calcined FMS.

Next, the effect of different calcination temperatures (all higher than 500 °C) on crystallinity is investigated. Fig. 3 shows the diffraction patterns for the ceramic nanofibers calcined at increasing temperatures. No peaks associated with crystallographic phases are observed for materials calcined at temperatures lower than 538 °C. For the sample calcined at this temperature, initial formation of the anatase phase of titanium oxide can be observed. A peak corresponding to the diffraction plane (101) is identified. In contrast, at higher temperatures (e.g., 550 °C), the occurrence of anatase titanium oxide (diffraction plane (101)) and fluorite cerium oxide (diffraction plane (111)) phases is detected. An effect of cerium dopant in inhibiting the formation of anatase phase is observed, which is supported by literature.^21–23,35 This effect is likely due to large dimensional cation mismatch (Ce⁴⁺ (0.87 Å) vs. Ti⁴⁺ (0.60 Å) in 6-fold coordination) that, at lower temperatures (<550 °C), causes reduced elemental diffusion of the cations. By contrast, at temperatures higher than 550 °C, elemental diffusion is efficiently activated and the Ce-saturated TiO₂ anatase phase crystallizes.²²

Fig. 3 XRD patterns of Ce–W–TiO₂ FSM calcined at different temperatures. The fluorite phase of cerium oxide is indicated with a star.

Based on XRD patterns, further characterizations are only conducted on three selected samples: FSM@500, FSM@538, FSM@575; the three samples are selected as representatives of an amorphous state, an initial stage of phase formation, and a crystallographic phase, respectively.

The effect of calcination temperature on morphology is investigated by SEM and TEM (Fig. 4a–f) for the three selected samples. The micrographs refer to membranes previously cut into discs with diameter of around 15 mm after calcination (insets in Fig. 4b, d, and f). SEM images (Fig. 4a, c, and e) indicate defect-free nano-textured porous structures composed of nanofibers with high aspect ratios and narrow diameter distribution. Open and interconnected porosity is also observed and associated with a large porosity volume. A slight contraction of the diameter of nanofibers upon increasing the calcination temperature is observed and explained as an effect of the temperature (Table 1). Fig. 4b, d, and f show the corresponding TEM micrographs for the above discussed samples. In agreement with the XRD patterns, amorphous characteristics, corresponding to a smooth surface, are observed for the nanofibrous membranes formed when calcined at 500 °C (Fig. 4b). In the corresponding diffraction pattern obtained using a selected-area-electron diffraction (SAED) technique (inset in Fig. 4b), no signals for an ordered disposition of atoms within a crystallographic unit cell are observed. Instead, the initial formation of crystallites on the surface of nanofibers can be inferred for the membrane calcined at 538 °C (Fig. 4d). The incipient formation of a crystallographic phase is also supported by the corresponding SAED pattern (inset in Fig. 4d), where an early signal of the crystallographic phase can be identified. Eventually, characteristically rough surfaces and densely packed polycrystalline morphology are observed for the nanofibrous membranes calcined at 575 °C (Fig. 4f), which is confirmed by the corresponding SAED (inset in Fig. 4f).

Fig. 4 Top-view SEM micrographs of Ce–W–TiO_2−δ FSM calcined at (a) 500 °C, (c) 538 °C, and (e) 575 °C; TEM images of Ce–W–TiO_2−δ FSM calcined at different temperatures: (b) 500 °C, (d) 538 °C, and (f) 575 °C; the lower-insets in (b), (d), and (f) are the corresponding selective area electron diffraction (SAED) patterns and the upper-insets are the digital photos of Ce–W–TiO₂ FSM discs.

Table 1 Characteristics of the nanofibers (phase, diameter, specific surface area, pore diameter, pore volume) calcined at 500, 538, and 575 °C

Sample	Crystallographic phase	Average nanofiber diameter (nm)	Specific surface area (m^2 g^−1)	Pore volume (cm^3 g^−1)	Pore diameter (nm)
CeWTiO2@500	Amorphous	229 ± 42	59.89	0.08	3.5
CeWTiO2@538	Anatase/fluorite	219 ± 53	43.48	0.13	7.3
CeWTiO2@575	Anatase/fluorite	212 ± 59	38.88	0.13	9.1

---

The surface properties of the nanofibers, which are of critical relevance for understanding the catalytic performances, are also investigated. Fig. 5a shows the N₂ adsorption–desorption curves, whereas pore volume, pore diameter, and specific surface area (SSA) data are summarized in Table 1. Adsorption curves for the three membranes follow a typical V type profile of mesoporous materials (pore diameter: 2–50 nm) with an H2 hysteresis loop. Interestingly, while the surface pore volume increases from the sample calcined at 500 °C to the sample calcined at 538 °C, it remains constant when the temperature is increased to 575 °C (Fig. 5b and Table 1). Continuous increment of the pore diameter values with temperature is instead observed (Fig. 5b and Table 1), which corresponds to decrease in the specific surface area (Table 1). The trends for surface parameters are observed due to temperature, which is expected to induce the growth of TiO₂ crystallites.³⁶

Fig. 5 (a) N₂ adsorption–desorption isotherms of Ce–W–TiO_2−δ FSM calcined at 500 °C, 538 °C, and 575 °C; (b) pore size distributions of the different FSM catalysts. The inset in (b) shows their corresponding pore diameters and pore volumes.

Next, the structural characteristics of the calcined FSM are investigated with the aim of achieving an insight into the porosity. Details on the porous membrane structure can be obtained via FIB/SEM analysis. Fig. 6a shows the SEM image of the membrane calcined at 500 °C after FIB slicing, which is recorded at high magnification with the In-lens detector at 1 kV. Inside the hole, dug with a gallium probe, two phases are observed: the white textural structure, which corresponds to TiO₂ nanofibers, and the black bulk, which is the epoxy resin used to infiltrate the pores. 3D tomography reconstruction of sub-volume of the fiber-based membranes (7 × 15 × 8 μm³) is shown in Fig. 6b. The TiO₂ fibers are represented in white, whereas the pores are shown as transparent. High porosity of the developed membrane is undeniably evident form the 3D reconstruction image. An analysis of the image provides an estimated value of around 96%; this is an extremely high value, and it is larger than the value of ca. 80% (ref. 21) reported in literature for a nanofibrous catalyst and that obtained by analyzing SEM micrographs (via estimation of the mean intercept length).³⁷ The high porosity obtained for the nanofibrous FSM is likely due to PVA, which not only prevents excessive bending of the single nanofibers, but also limits shrinkage of the porous membrane during the thermal removal of alkoxides and PMMA. Further relevant details on both the porous network structure and nanofibers are shown in Fig. 7a and b, respectively. Fig. 7a indicates average pore diameter of around 650 nm, whereas Fig. 7b indicates average fiber length of around 12.3 μm. It is worth mentioning that the latter value is influenced by the dimension of the volume analyzed, and its value is expected to be higher when the entire volume of the membrane is considered.

Fig. 6 (a) SEM micrograph of FSM calcined at 500 °C after FIB slicing, and (b) the corresponding 3D FIB tomography reconstruction of a sub-volume (7 × 15 × 8 μm³).

Fig. 7 (a) The pore size distribution and (b) the fiber length distribution corresponding to FSM calcined at 500 °C.

The catalytic activity of the developed membranes is then explored, and the obtained information is used to better understand the performances. The aim of the study is characterization of the intrinsic catalytic activity of highly porous FSM. Accordingly, experiments are conducted in dry conditions and in the absence of SO₂. FSMs in the shape of discs with diameter of 15 mm (after calcination) and thickness of approximately 100 μm are tested. Each catalytic disc corresponds to an extremely low weight of material (CeWTiO₂) of around 8 mg.

Fig. 8a shows NO_x conversion as a function of reaction temperature for the three membranes FSM@500, FSM@538, and FSM@575. Three regions in the catalytic activity of the materials can be identified: low, intermediate, and high temperatures. In the first region, the NO_x conversion values are low and almost constant, whereas a steady increase is observed in the intermediate range of temperatures until a plateau, corresponding to the maximum NO_x conversion at approximately 300 °C. This represents the regime region, where the reaction is only controlled by the transport of reagents. Interestingly, for all three investigated catalysts, NO_x conversion declines at temperature as high as 500 °C, which is likely due to the increased rate of ammonia oxidation at these temperatures.³⁸ In the intermediate temperature region, differences in NO_x conversion trends for the three samples are observed; amorphous FSM (FSM@500) is the best performing material, and this is characterized by conversion values of around 60% and 90% at 200 °C and 250 °C, respectively. Details on the reaction mechanism are obtained by exploring kinetics of the process.³⁹ Estimation of the activation energy (E_a) and of the number of active sites associated with the pre-exponential factor (A) can be obtained by the Arrhenius fitting of the experimental data. Fig. 8b shows Arrhenius plots for the three catalytic membranes, whereas the corresponding fitting parameters together with T_a are reported in Table 2. In the region at the lowest temperatures, Arrhenius analysis is not conducted because of the insufficient number of experimental data (only two). At the intermediate range, the plot indicates a reaction dependent on temperatures, whereas at the high-temperature region, no dependence is observed with activation energy values (E_a) close to zero. The transition from the intermediate to the regime region shifts to higher values with increased calcination temperature of the catalytic membranes (inset in Fig. 8b). The observed trend follows the order FSM@500 < FSM@538 < FSM@575. The amorphous catalyst has the lowest activation temperature (259 °C) and thereby, it is characterized by the widest regime region. Interestingly, regardless of the calcination temperature and the crystallographic phase, similar E_a values are observed, suggesting similar reaction mechanisms for the three catalytic FSMs. In contrast, a significant difference in the pre-exponential factors (A) is obtained. Specifically, similar numbers of active sites (within experimental errors) are observed for FSM@500 and FSM@538, whereas the lowest number of active sites is associated with FSM@575, which is likely due to the highest calcination temperature. Overall, superior catalytic performance is obtained for FSM@500, which is possibly due to the amorphous phase and correspondingly larger specific surface area (Table 1). Indeed, the amorphous phase is generally regarded as promising for a substrate for catalysis as it is considered chemically homogeneous with respect to the related surface properties.^22,40,41

Fig. 8 (a) SCR activity of Ce–W–TiO_2−δ FSM calcined at 500 °C, 538 °C, and 575 °C and (b) the corresponding Arrhenius plots.

Table 2 Arrhenius fitting parameters for the investigated materials (intermediate range of temperatures)

Sample	Activation energy, Ea (kJ mol^−1)	Pre-exponential factor, A	Activation temperature, Ta (°C)
CeWTiO2@500	40.10 ± 3.60	6.16 × 10^6 ± 0.15 × 10^6	259
CeWTiO2@538	41.74 ± 2.15	7.34 × 10^6 ± 0.23 × 10^6	274
CeWTiO2@575	40.38 ± 2.28	2.87 × 10^6 ± 0.11 × 10^6	300

---

The catalytic performance of FSM@500 is further explored. In Fig. 9a, the NO_x conversion values at different values of gas hourly space velocity (GHSV) are plotted as a function of temperature. As expected, high conversion (∼98%) is obtained at the lowest values of GHSV (163 000 h⁻¹ and 326 000 h⁻¹) within a relatively wide range of temperature (300–450 °C). Notably, at GHSV of 653 000 h⁻¹, the catalyst still retains an NO_x conversion value of around 90%. Fig. 9b shows the comparison for NO_x conversions by CeWTiO₂ nanofibers and by the state-of-the-art V–W–TiO₂ nanofibers and nanoparticles (as provided by Haldor Topsøe, Denmark). For clarity, the reference materials are hereafter indicated as CeWTiO₂-NF, V–W–TiO₂-NF, and V–W–TiO₂-PWD HD. The nanofibrous FSM undoubtedly exhibits the highest conversion values in the entire range of investigated temperatures and at different GHSV experimental conditions. This can be ascribed to the extended and highly porous network, which is characteristic of nanofibrous FSM and enables a favourable fluid-dynamic regime through the membrane. In such a structure, any single nanofiber can be considered as a nanoreactor arranged with other nanofibers into a highly porous and lightweight SCR unit. In Table 3, a comparison between NO_x conversions of nanofibrous FSM developed in the current study and literature data from both ceria-based mixed oxide nanofibers and nanoparticles^22,42–44 is displayed. The comparison is only reported at temperatures lower than 300 °C (150 and 200 °C) before the catalysts reach the regime region. Superior catalytic behavior of nanofibrous FSMs is observed over both the nanofibers and nanoparticles. Indeed, conversion values of around 30% and 73% are obtained at 150 °C and 200 °C, respectively, for FSM at GHSV of 163 000 h⁻¹; at the same temperatures, the literature data indicate conversion values of around 15% and 40–60%, respectively, at any GHSV values (e.g., 25 000, 200 000, and 250 000 h⁻¹).^22,42–44

Fig. 9 (a) NO_x conversions at different GHSVs by Ce–W–TiO_2−δ FSM calcined at 500 °C and (b) the comparison of NO_x conversion of same-composition nanofibers with that of the state-of-the-art material (V–W–TiO₂) in the form of nanofibers and nanoparticles. The latter catalyst is kindly provided by Haldor Tpsøe A/S.

Table 3 Literature data for ceria–tungsten-based mixed oxide catalysts. The compositions are reported indicating the Ce/W molar ratio and considering equal to the unit (1) molar content of Ti. For each reference, only the data of the best performing catalysts are listed

Catalyst	[NH3][NO] (ppm)	GHSV (h^−1)	NOx conv. (%) @150 °C	NOx conv. (%) @200 °C	Ref.
Ce1W1TiOx powder	500	250 000	10	50	38
Ce1W1TiOx powder	500	25 000	15	60	39
Ce3W1TiOx powder	1000	200 000	<10	40	40
Ce2.6W1TiOx Nanofibers	1000	200 000	19	56	22
Ce2.6W1TiOx nanofibers	1000	400 000	16	51	22
Ce2.6W1TiOx nanofibrous membranes (FSM@538)	1000	163 000	30	73	Current study
Ce2.6W1TiOx nanofibrous membranes (FSM@538)	1000	326 000	24	64	Current study

---

Long-term stability of the FSM@500 sample is also investigated over time frames of 120 h, at 300 °C, and at 326 000 h⁻¹ (Fig. 10a). Remarkably, the material exhibits a stable and high de-NO_x conversion value over the entire range of time frame investigated. No loss of activity is observed for the catalysts after a continuous long-term stability test (inset) compared to that for fresh materials. The stability of the material after 5 catalytic cycles is also verified (Fig. 10b). The reproducible conversion profiles observed further confirm the high stability of the developed FSM at working conditions. Notably, FSM@500 disc still retains its integrity after 5 SCR testing cycles (Fig. 10c).

Fig. 10 (a) Long-term stability test (120 h) at 300 °C of FSM@500 and (b) its reproducibility test over 5 working cycles. Both tests are performed at GHSV of 326 000 h⁻¹. (c) Digital photos of fresh FSM@500 and after 5 SCR testing cycles.

Conclusions

Free-standing Ce–W–TiO_2−δ nanofibrous membranes (FSM) are successfully electrospun to serve as catalysts for the removal of NO_x in a selective catalytic reduction (SCR) process. Structural stability is achieved by integrating into the nanofibrous membrane a co-electrospun polymeric component, which can stabilize the catalytic nanofibers during the removal of organic additives at around 300 °C. The supporting polymer used is PVA, and its complete thermal decomposition occurs at around 500 °C. In such a combination, PVA can withstand structural stress due to the removal of additives and the corresponding shrinkage. PVA is also proven to limit the volumetric contraction of the whole membrane. This results in an extremely high porosity value of ca. 96%.

The surface and morphological features of the nanofibers along with structural and porosity characteristics of the resulting FSM are definitely the main factors responsible for the exceptional catalytic performances. Indeed, these characteristics are expected to favor a fluid-dynamic regime that enables efficient exposure of the active sites to the reactant gases, thereby enabling high NO_x conversion levels: values as high as 95% at GHSV of 326 000 h⁻¹ are observed over a wide range of temperatures. Moreover, high conversion values (ca. 90%) are maintained even at high gas speeds (653 000 h⁻¹). Notably, a comparison with nanofibers (NFs) of the same composition indicates superior performances for nanofibrous FSM. Besides, high long-term stability over multiple working cycles is also demonstrated.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors would like to express their gratitude for financial supports from the Royal Golden Jubilee (RGJ) PhD Program within Thailand Research Fund (TRF), (Contract No. PHD/0046/2556).

References

1. H. Liang, X. Jiao, C. Li and D. Chen, Flexible self-supported metal-organic framework mats with exceptionally high porosity for enhanced separation and catalysis, J. Mater. Chem. A, 2018, 6, 334–341.
2. P. Marchetti, M. F. J. Solomon, G. Szekely and A. G. Livingston, Molecular Separation with Organic Solvent Nanofiltration: A Critical Review, Chem. Rev., 2014, 114, 10735–10806.
3. Q. Wen, J. Di, Y. Zhao, Y. Wang, L. Jiang and J. Yu, Flexible inorganic nanofibrous membranes with hierarchical porosity for efficient water purification, Chem. Sci., 2013, 4, 4378–4382.
4. Y. Wang, W. Li, Y. Xia, X. Jiao and D. Chen, Electrospun flexible self-standing γ-alumina fibrous membranes and their potential as high efficiency and fine particulate filtration media, J. Mater. Chem. A, 2014, 2, 15124–15131.
5. E. Y. S. Tan, S. Agarwala, Y. L. Yap, C. S. H. Tan, A. Laude and W. Y. Yeong, Novel method for the fabrication of ultrathin, free-standing, and porous membranes for retinal tissue engineering, J. Mater. Chem. B, 2017, 5, 5616–5622.
6. W. Kujawski, J. Kujawa, E. Wierzbowska, S. Cerneaux, M. Bryjak and J. Kujawski, Influence of hydrophobization conditions and ceramic membranes pore size on their properties in vacuum membrane distillation of water–organic solvent mixtures, J. Membr. Sci., 2016, 499, 442–451.
7. D. Wu, F. Xu, B. Sun, R. Fu, H. He and K. Matyjaszewski, Design and Preparation of Porous Polymers, Chem. Rev., 2012, 112, 3959–4015.
8. B. A. Pulido, C. Waldron, M. G. Zolotukhin and S. P. Nunes, Porous polymeric membranes with thermal and solvent resistance, J. Membr. Sci., 2017, 539, 187–196.
9. J. A. Lewis, Colloidal processing of ceramics, J. Am. Ceram. Soc., 2000, 83, 2341–2359.
10. D. Marani, J. Hjelm and M. Wandel, Rheological analysis of stabilized cerium-gadolinium oxide (CGO) dispersons, J. Eur. Ceram. Soc., 2014, 34, 695–702.
11. D. Marani, C. Gadea, J. Hjelm, P. Hjalmarsson, M. Wandel and R. Kiebach, Influence of hydroxyl groups of binders on rheological properties of cerium-gadolinium oxide (CGO) screen printing inks, J. Eur. Ceram. Soc., 2015, 35, 1495–1504.
12. D. Marani, V. Esposito, B. R. Sudireddy, J. J. Bentzen, P. S. Jørgensen, D. W. Ni, F. Teocoli and R. Kiebach, Enhanced densification of thin tape cast ceria-gadolinium oxide (CGO) layers by rheological optimization of slurries, Ceram. Int., 2017, 43, 5647–5653.
13. F. Teocoli, D. Marani, R. Kiebach and V. Esposito, Effect of spherical porosity on co-fired dense/porous zirconia bi-layers cambering, J. Eur. Ceram. Soc., 2018, 38, 173–179.
14. C. G. Schmidt, K. K. Hansen, K. B. Andersen, Z. Fu, A. Roosen and A. Kaiser, Effect of pore formers on properties of tape cast porous sheets for electrochemical flue gas purification, J. Eur. Ceram. Soc., 2016, 36, 645–653.
15. S. Thenmozhi, N. Dharmaraj, K. Kadirvelu and H. Y. Kim, Electrospun nanofibers: new generation of materials for advanced applications, Mater. Sci. Eng., B, 2017, 217, 39–48.
16. W. Li, Y. Wang, B. Ji, X. Jiao and D. Chen, Flexible Pd/CeO₂-TiO₂ nanofibrous membranes with high efficiency ultrafine particulate filtration and improved CO catalytic oxidation performances, RSC Adv., 2015, 5, 58120–58127.
17. R. Sahay, P. Suresh Kumar, R. Sridhar, J. Sundaramurthy, J. Venugopal, S. G. Mhaisalkar and S. Ramakrishn, Electrospun composite nanofibers and their multifaceted applications, J. Mater. Chem., 2012, 22, 12953–12971.
18. F. Zhang, C. Yuan, J. Zhu, J. Wang, X. Zhang and X. W. Lou, Flexible films derived from Electrospun carbon nanofibers incorporated with Co₃O₄ hallow nanoparticles as self-supported electrodes for electrochemical capacitors, Adv. Funct. Mater., 2013, 23, 3909–3915.
19. M. Shang, W. Wang, S. Sun, E. Gao, Z. Zhang and L. Zhang, R. O'Hayre, The design and realization of a large area flexible nanofiber-based mat for pollutant degradation: an application in photocatalysis, Nanoscale, 2013, 5, 5036–5042.
20. Y. Zhu, X. Han, Y. Xu, Y. Liu, S. Zheng, K. Xu, L. Hu and C. Wang, Electrospun Sb/C fibers for a stable and fast sodium-ion battery anode, ACS Nano, 2013, 7, 6378–6386.
21. D. Marani, R. H. Silva, A. Dankeaw, K. Norrman, R. M. L. Werchmeister, D. Ippolito, M. Gudik-Sørensen, K. K. Hansen and V. Esposito, NOx selective reduction catalysis (SCR) on self-supported V-W-doped TiO₂ nanofibers, New J. Chem., 2017, 41, 3466–3472.
22. A. Dankeaw, F. Gualandris, R. H- Silva, K. Norrman, M. Gudik-Sørensen, K. K. Hansen, B. Ksapabutr, V. Esposito and D. Marani, Amorphous saturated cerium–tungsten–titanium oxide nanofiber catalysts for NOx selective catalytic reaction, New J. Chem., 2018, 42, 9501–9509.
23. D. Marani, R. H. Silva, A. Dankeaw, M. Gudik-Sørensen, K. Norrman, K. K. Hansen and V. Esposito, Effect of the sol-gel conditions on the morphology and SCR performance of electrospun V-W-TiO2 catalysts, J. Phys. Chem. Solids, 2018, 118, 255–261.
24. L. Shahreen, G. G. Chase, A. J. Turinske, S. A. Nelson and N. Stojilovic, NO decomposition by CO over Pd catalyst supported on TiO₂ nanofibers, Chem. Eng. J., 2013, 225, 340–349.
25. S. Zhan, M. Qiu, S. Yang, D. Zhu, H. Yu and Y. Li, Facile preparation of MnO₂ doped Fe₂O₃ hollow nanofibersfor low temperature SCR of NO with NH₃, J. Mater. Chem. A, 2014, 2, 20486–20493.
26. S. G. Jennings, The mean free path in air, J. Aerosol Sci., 1988, 19, 159–166.
27. Y. Dai, W. Liu, E. Formo, V. Sun and Y. Xia, Ceramic nanofibers fabricated by electrospinning and their applications in catalysis, environmental science, and energy technology, Polym. Adv. Technol., 2011, 22, 326–338.
28. S. J. Park, G. G. Chase, K. U. Jeong and H. Y. Kim, Mechanical properties of titania nanofiber mats fabricated by electrospinning of sol–gel precursor, J. Sol-Gel Sci. Technol., 2010, 54, 188–194.
29. J. Sauvola and M. Pietikäinen, Adaptive document image binarization, Pattern Recognit., 2000, 33, 225–236.
30. T. Kryjak and M. Gorgoń, Parallel implementation of local thresholding in Mitrion-C, Int. J. Appl. Math. Comput. Sci., 2010, 20, 571–580.
31. B. Münch, Holzer, Contradicting geometrical concepts in pore size analysis attained with electron microscopy and mercury intrusion, J. Am. Ceram. Soc., 2008, 91, 4059–4067.
32. S. Pandey, S. K. Pandey, V. Parashar, G. K. Mehrotra and A. C. Pandey, Ag/PVA nanocomposites: optical and thermal dimensions, J. Mater. Chem., 2011, 21, 17154–17159.
33. S. Peng, M. Zhou, F. Liu, C. Zhang, X. Liu, L. Zou and J. Chen, Flame-retardant polyvinyl alcohol membrane with high transparency based on a reactive phosphorus-containing compound, R. Soc. Open Sci., 2018, 4, 170512.
34. R. E. Mistler and E. R. Twiname, in Tape Casting: Theory and practice, ed. John Wiley and Sons, Inc., 2006.
35. M. Reli, N. Ambrožova, M. Šihor, L. Matějová, L. Čapek, L. Obalová, Z. Matě, A. Kotarba and K. Koci, Novel cerium doped titania catalysts for photocatalytic decomposition of ammonia, Appl. Catal., B, 2015, 178, 108–116.
36. V. Esposito, D. Marani, F. Teocoli, K. Tor Sune Thydén, D. Z. De Florio and F. C. Fonseca, Accelerated ceria–zirconia solubilization by cationic diffusion inversion at low oxygen activity, J. Mater. Chem., 2016, 4, 16871–16878.
37. M. M. Tomadakis and T. J. Robertson, Viscous permeability of random fiber structures: comparison of electrical and diffusional estimates with experimental and analytical results, J. Compos. Mater., 2005, 39, 163–188.
38. S. S. Reddy Putluru, L. Schill, D. Gardini, S. Mossin, J. B. Wagner, A. D. Jensen and R. Fehrmann, Superior DeNOx activity of V₂O₅–WO₃/TiO₂ catalysts prepared by deposition–precipitation method, J. Mater. Sci., 2014, 49, 2705–2713.
39. G. Dong, Y. Bai, Y. F. Zhang and Y. Zhao, Effect of the V^4+(3+)/V5+ ratio on the denitration activity for V₂O₅–WO₃/TiO₂ catalysts, New J. Chem., 2015, 39, 3588–3596.
40. C. Yoon and D. L. Cocke, Potential of amorphous materials as catalysts, J. Non-Cryst. Solids, 1986, 39, 217–245.
41. X. Tang, J. Hao, W. Xu and J. Li, Low temperature selective catalytic reduction of NOx with NH3 over amorphous MnOx catalysts prepared by three methods, Catal. Commun., 2007, 8, 329–334.
42. W. Shan, F. Liu, H. He, X. Shi and C. Zhang, A superior Ce-W-Ti mixed oxide catalyst for the selective catalytic reduction of NOx with NH₃, Appl. Catal., B, 2012, 115–116, 100–106.
43. K. Zhao, W. Han, G. Lu, J. Lu, Z. Tang and X. Zhen, Promotion of redox and stability features of doped Ce–W–Ti for NH3-SCR reaction over a wide temperature range, Appl. Surf. Sci., 2016, 379, 316–322.
44. Y. Jiang, Z. Xing, X. Wang, S. Huang, X. Wang and Q. Liu, Activity and characterization of a Ce–W–Ti oxide catalyst prepared by a single step sol–gel method for selective catalytic reduction of NO with NH₃, Fuel, 2015, 151, 124–129.

---