Effect of pore-directing agents and silanol groups in mesoporous silica nanoparticles as Nafion fillers on the performance of DMFCs

Abstract

Two kinds of nanoparticles of mesoporous silica, namely SBA-15n and MSN, were prepared using P123 and CTMABr as pore-directing agents, respectively, and loaded into Nafion® to form composite membranes by solvent casting method. The physico-chemical properties of these nanoparticles were examined with powder-XRD, N₂ sorption, TGA, EA, SEM and ²⁹Si MAS NMR. The methanol permeability, proton conductivity, and cell performance of the resultant composite membranes were compared in terms of the amount of nanoparticles, whether the pore-directing agents were removed and the different kinds of nanoparticles of mesoporous silica. It was found that both pore-directing agents present inside the pores of nanoparticles could contribute to resisting methanol crossover to the cathode. However, only P123 inside SBA-15n could assist proton transfer, probably through the ether groups on the P123 co-polymer. The proton conductivity of the composite membrane containing extracted SBA-15n was lower than the membrane that contained P123. Nevertheless, the highest proton conductivity was obtained on the composite membranes filled with ethanol extracted MSN (Ex-MSN) particles. This is attributed to the larger amount of silanol groups present in Ex-MSN than in extracted SBA-15n. The optimal proton conductivity of 4.01 × 10⁻² S cm⁻¹ was obtained on the composite membrane filled with 5% Ex-MSN, and the single cell assembled with this composite membrane gave a highest power density of 131 mW cm⁻² at 60 °C, which was about 2 times higher than the cell with recasting plain Nafion membrane and 36% higher than that with commercial Nafion® 117 membrane.

---

1. Introduction

Among various fuel cells, the direct methanol fuel cell (DMFC) is a promising power source for various applications, especially for portable devices or transportation owing to its high energy density at low operating temperature and easy storage of the fuel. However, slow oxidation kinetics of methanol at the anode and methanol crossover from the anode to the cathode are the major drawbacks for the commercialization of DMFCs.^1–5 The polymer electrolyte membrane, which is employed to provide proton conduction from the anode to the cathode and effective separation of the methanol fuel at the anode and oxygen at the cathode, needs to work in high humidity and therefore also conduces methanol crossover. Perfluorosulfonic acid (PFSA) membranes such as Nafion (DuPont) are the most widely used electrolyte material in DMFCs due to their high thermal and chemical stabilities as well as excellent mechanical strength.^6–9 However, high methanol permeation through PFSA polymer membranes significantly lowers fuel efficiency and cell performance, and thus impedes the commercial development of DMFCs.

There have been attempt to reduce the methanol permeability through the PFSA membranes by making a composite with inorganic materials, which introduce a winding pathway for the methanol molecules. This approach has attracted a large group of researchers,^10–29 and some efforts have been put in studying the composite membranes of Nafion for DMFCs by incorporating silica,^10–15 titania,^16–18 zirconia,^19–21 zeolites,²² heteropolyacids,^23–26 and zirconium phosphates^26–29 in the membrane. These composite membranes did restrict methanol crossover, but the poor proton conductivities limited their practical applications in DMFCs.

Recently, various types of mesoporous silica materials with ordered pores have been studied due to their high surface areas and well-ordered porous structures.^30–32 They were investigated as high-temperature/low-humidity proton conductors owing to their excellent thermal stabilities and water adsorption capabilities.^33,34 Mesoporous silica materials of different porous structures functionalized with sulfonic acid groups were pressed to pellets and their proton conductivities were compared by Marschall et al.³⁴ They found that SBA-16, which contains three dimensional (3D) cage-like porous structures, gave lower proton conductivities than the mesoporous materials containing one dimensional (1D) channeling pores, saying SBA-15 and MCM-41, attributing to that the non-directing 3D pore arrangement cannot guide the protons and water molecules as good as the 1D channeling pore arrangement. Jiang et al.³⁵ prepared pellets of sulfonic acid functionalized mesoporous silica with the aid of polyvinylidene fluoride adhesive. Three periodic ordered mesoporous structures were compared, including 2D hexagonal, 3D body-centered cubic and 3D cubic bicontinuous structures. Among them, the 2D hexagonal materials containing 1D channeling pores showed higher proton conductivity than the other two mesoporous structures. Although the silica membranes have the advantage of high operation temperatures (120–200 °C), the proton conductivities of pure silica membranes functionalized with sulfonic acid groups are relatively low (ca. 10⁻³ S cm⁻¹ at 140 °C)³⁴ After mixing with polyvinylidene fluoride adhesive, the proton conductivities were reported to greatly enhance.³⁵ However, the performances of DMFC assembled with mesoporous silica pellets as the proton exchange electrolyte were not shown.

Better proton conductivities and mechanical properties are usually obtained when mesoporous silica materials are loaded in polymeric membranes as inorganic fillers.^36–42 Furthermore, the proton conductivities, water uptake and water retention can be enhanced when protic functional groups such as sulfonic acid and amino groups are incorporated onto mesoporous silica.^35,43–52 Using mesoporous silica as an inorganic filler to form composite membranes was first reported by Tominaga et al.⁴² in 2007. They found that addition of mesoporous silica SBA-15 could increase proton conductivity of Nafion matrix due to its large surface area. However, the composite membrane was only applied to hydrogen fuel cell and not for DMFC. In 2008, Jin et al.³⁶ developed composite membranes by incorporating MCM-41 nanospheres into Nafion membranes, and found that the MCM-41 nanoparticles aggregated in the membranes if the loading was up to 3 wt%. The size of MCM-41 nanospheres was about 50 nm, similar to mesoporous silica nanoparticles (MSN) reported in the literature.^53–56 The DMFC assembled with the composite membrane exhibited slightly higher power density than recasting Nafion (21.8 mW cm⁻² vs. 18.4 mW cm⁻²), while no comparison was made with commercial Nafion membrane.

The methanol permeability of the composite membrane has been related to the size of the filler particles. Byun et al.⁵⁷ prepared Nafion membranes loaded with ZSM-5 zeolite of different sizes and forms, and they found that small ZSM-5 particles (200–300 nm) mainly existed within the ionic cluster channels in the composite membranes could effectively block methanol crossover from anode to cathode. In contrast, the large sized zeolites of ca. 1 μm showed higher methanol permeabilities than those of 200–300 nm due to that methanol could transport through void space between large zeolite particles and Nafion.

SBA-15 and MCM-41 materials of two-dimensional channeling pores arranged in hexagonal P6mm structure have received great attention because of their relatively large pores and high hydrothermal stabilities. Conventional SBA-15 is synthesized with TEOS as the silica source and P123 as pore directing agent at a relatively strong acidic condition,³⁰ while conventional MCM-41 is synthesized using CTMABr as pore directing agent and sodium silicate or TEOS as the silica source at a relatively strong basic condition.³¹ However, the sizes of conventional SBA-15 and MCM-41 particles are usually in micrometers, which is too big to fill in the water channels of Nafion membrane. In the past decade, nanoparticles of MCM-41 has been prepared, namely “mesoporous silica nanoparticles” (MSN), and applied to drug delivery,^58,59 cell tracking^60,61 and catalysis.^62,63 Incorporating MSN as inorganic fillers into Nafion membranes was first reported by Jin et al.³⁶ Enhancement of water retention, improvement in thermal stability, and reduction of methanol crossover were observed in comparison to those of pristine Nafion. On the other hand, conventional SBA-15 particles have also been added in polymer membranes (Nafion, SPEEK, polystyrene-block-poly(ethylene-ran-butylene)-block-polystyrene and polybenzimidazoles) as fillers. However, most of them were applied in H₂-PEMFC.^{45,46,52,64–68} To the best of our knowledge, only one paper reported by Cho et al.⁶⁹ loaded conventional SBA-15 as fillers in polymer membranes and applied to DMFC. They compared phenyl-sulfonic acid functionalized mesoporous silica materials, saying SBA-15 and MCM-41, as inorganic fillers in poly(vinyl alcohol)-based membranes and found that methanol permeability of the composite membranes increased with the increase of pore size, attributing to enhanced diffusivity of methanol in large pores. Nevertheless, none of previous literatures has examined the effects of pore-direct agents in the mesopores and amounts of silanol groups.

In this study, SBA-15 nanoparticles (SBA-15n) of about 100–300 nm were prepared by a novel method developed in our laboratory. The SBA-15n and MSN of about 100 nm were used as the inorganic fillers of Nafion membrane. Nanosized particles are expected to provide a homogenous distribution as the fillers in the Nafion matrix. On the other hand, the pores of mesoporous silica are bigger than methanol molecule (0.41 nm), so the methanol may transfer through mesopores from cathode to anode. Therefore, the effect of pore-directing agents in the mesopores of silica nanoparticles on methanol permeability, proton conductivity, and DMFC performance were also examined. The pore-directing agent of SBA-15n is P123, which is a triblock co-polymer of ethylene glycol and propylene glycol units, while that of MSN is CTMA⁺, a quaternary ammonium cation surfactant. Furthermore, the silanol groups of extracted SBA-15n and MSN were characterized with solid-state ²⁹Si NMR, and the effect of the amount of silanol groups on the mesoporous silica on the proton conductivities of the composite membranes was studied.

2. Experimental

2.1. Materials

Nafion solution D520 (EW = 1000) from DuPont™ with polymer content of 5 wt% in 1-propanol was used to prepare composite membranes. Triblock copolymer, poly(ethylene glycol)–poly(propylene glycol)–poly(ethylene glycol) (Pluronic P123, formulated as EO₂₀PO₇₀EO₂₀, Aldrich M_n = 5800) and cetyltrimethylammonium bromide (CTMABr, Acros 99+%) were used as the pore-directing agents. Sodium silicate solution (Aldrich, SiO₂ ∼ 27%) and tetraethyl orthosilicate (TEOS, Seedchem 98%) were used as the silica sources. N,N-Dimethylacetamide (DMAc, Acros 99%), acetone (Merck 99.5%), methanol (Echo 99.9%), normal butanol (n-BuOH, Acros 99%), sulfuric acid (Schalau 95–97%), hydrogen peroxide (Acros 35%) and hydrochloric acid (Aencore ∼36%) were purchased from commercial suppliers and used without further purification.

2.2. Synthesis of SBA-15 nanoparticles (SBA-15n) with/without surfactants

7.5 g of 2 M HCl_(aq) and 2 g of Pluronic P123 were dissolved in 40 g of D.I. water, and the solution was stirred at room temperature overnight. Then, the temperature of the solution was raised to 35 °C and the stirring was continued for another 3 h. 4.2 g of sodium silicate was then added dropwise into the above solution until the pH value of the solution reached 5.5. After stirring for 24 h, the mixture was sealed in a polypropylene bottle and hydrothermally heated at 90 °C for 24 h. The solid product was washed three times with H₂O and collected by centrifugation (11 000 rpm for 25 min) and then stored in DMAc solution. The as-made SBA-15 containing P123 surfactant as the pore-directing agent is denoted as S-SBA-15n, where “S” represents “surfactant” and “n” designates “nanoparticle”.

The pore-directing agent in S-SBA-15n was removed by extraction with ethanol at reflux temperature rather than by calcination, since the latter process would result in aggregation of SBA-15 nanoparticles. The S-SBA-15n was extracted two times with 500 mL of ethanol containing 1 mL of concentrated HCl_(aq), followed by washing twice with ethanol and collected by centrifugation. Then, the washed solid was stored in DMAc. The resultant material was denoted as Ex-SBA-15n, where “Ex” represents “extraction”.

2.3. Synthesis of MSN (mesoporous silica nanoparticles) with/without surfactants

The MSNs were synthesized according to the procedures reported in the literature.⁷⁰ Typically, 0.3968 g of CTMABr and 70 mL H₂O were mixed at 40 °C and stirred for 1 h. Then, 1.81 mL of 28–30% NH_3(aq) was added. After stirring for 1 h, 1.74 mL of TEOS was added and the mixture was violently stirred at room temperature for 6 h. The solid product was washed three times with H₂O and collected by centrifugation (11 000 rpm for 25 min), and then stored in DMAc solution. The as-made MSN contains CTMA⁺ surfactant as the pore-directing agent is denoted as S-MSN, where “S” represents “surfactant”. The pore-directing agent in MSN was also removed by extraction with ethanol at reflux temperature rather than by calcination to avoid aggregation of MSN. The washed MSN was denoted as Ex-MSN, where “Ex” represents “extraction”, and it was also stored in DMAc solvent.

2.4. Preparation of composite membranes

18 g of Nafion® solution was placed in a vial and dried at 90 °C for 1 day. Then the appropriate amount of Ex-SBA-15n (or Ex-MSN), DMAc (the total amount of DMAc is 4.5 mL) and 1.5 mL acetone were added and stirred at room temperature. This solution was stirred for 1 day and ultrasonicated for 3 h to ensure good mixing. Finally, 5 mL mixture solution was poured into a Petri dish. The composite membranes were obtained after the solvent in the dish was evaporated in an oven at 90 °C for 48 h and then 120 °C for 12 h. In order to compare with composite membranes, the recasting Nafion membrane was prepared in the similar procedure without the addition of Ex-SBA-15n or Ex-MSN.

All composite membranes and recasting Nafion membrane were boiled first in 6 wt% H₂O₂ for 1 h, then in DI water for another hour, followed by in 0.5 M H₂SO₄ for 1 h. Then, the membranes were rinsed thoroughly with DI water and immersed in DI water for 24 h to remove residual acids. Finally, these membranes were stored in DI water. The as-prepared composite membranes were denoted as x%-Ex-SBA-15n or x%-Ex-MSN, where x refers to the weight percentage of SBA-15n or MSN material relative to Nafion®. After dried at room temperature, thicknesses of as-prepared membranes were measured, and the data are shown in Table S1 (ESI†). The thicknesses are all around 200 μm.

2.5. Characterization of filler and membranes

Powder X-ray diffraction (XRD) patterns were recorded in the 2θ range of 0.5–5° by a PANalytical X'pert Pro diffractometer with Cu Kα (λ = 1.5418 Å) radiation operated at 40 mA and 45 kV. N₂ sorption isotherms were measured using a Micromeritics Tristar 3000 system at liquid nitrogen temperature (77 K). Before the measurements, the samples were degassed at 373 K overnight under 10⁻³ Torr vacuum. The surface area was calculated by using Brunauer–Emmett–Teller (BET) method. Pore size was determined by Barrett–Joyner–Halenda (BJH) method using the desorption branch of the isotherms. The scanning electron microscopy (SEM) images and elemental mapping were taken using a Field Emission Scanning Electron Microscope with INCA X-Max EDS (JEOL JSM-7600F). Thermogravimetric analysis (TGA) was carried out in air using a Hi-Res TGA2950 instrument as well as a TA9530 thermometer. The SBA-15n or MSN were placed in a platinum holder and heated from room temperature to 50 °C at a heating rate of 10 °C min⁻¹ and maintained at 50 °C for 5 min before the weight losses were recorded up to 900 °C at a heating rate of 10 °C min⁻¹. For TG analyses of the composite membranes, the samples were held at 35 °C for 3 h before the weight losses were recorded up to 900 °C at a heating rate of 5 °C min⁻¹. The mechanical properties of membranes (5 × 50 mm) were obtained by using a DMA 2980 instrument (TA instruments) at 35 °C under N₂. The driving force for this apparatus was in the range between 0.0001 N and 18 N. SAXS (Small Angle X-ray Scattering) measurements were performed at beamline 23A SAXS at NSRRC (National Synchrotron Radiation Research Center) Taiwan. The wavelengths of X-ray is 1.24 Å.

2.6. Water uptake

The water uptakes of the membranes were measured by drying the membrane in a freeze dryer for 24 h before recording the weight. The dried membrane was immersed in distilled water for 24 h at room temperature. After wiping the wet membrane with lens tissue to remove excess water, the weight of membrane was recorded again. The difference in these two weights is water uptake of the membrane, as shown in eqn (1)where, W_wet and W_dry are the weights of the wetted and dried membranes, respectively.

2.7. Methanol permeability

Methanol permeability was measured using a manufactured permeation cell. It consists of two compartments (A and B) in identical volume separated by the membrane. Compartment A was filled with 68 mL aqueous solution of butanol (0.02 M) and methanol (2.00 M), while compartment B was only filled with 68 mL aqueous solution of butanol (0.02 M). The methanol permeability was determined by analyzing the methanol concentration in compartment B as a function of time using gas chromatography (Shimadzu GC-2014) equipped with a FID detector and a RTX-5 column. The methanol permeability (P, cm² s⁻¹) was calculated from eqn (2) and (3).where, dC_B/dt is the slope of the straight line of methanol concentration in compartment B versus time; V is the volume of aqueous solution; l and A are the thickness and the effective area of the membrane, respectively; C_A0 is the initial methanol concentration in compartment A; C_B0 is the initial methanol concentration in compartment B.

2.8. Proton conductivity

Proton conductivities of the membranes were measured by electrochemical impedance spectroscopy (EIS) using an impedance analyzer with a homemade two-point probe method at 90 °C and 90% RH in a humidity chamber. The impedance analyzer was operated in potentiostatic mode with voltage amplitude of 0.05 V at frequency range of 1000–10 kHz. The Nyquist plots of composite membranes are shown in Fig. S1 and S2, ESI.† At the points of Z′′ values equal to zero in Nyquist plots, the corresponding Z′ value is resistance (R). The proton conductivity σ (S cm⁻¹) was calculated from eqn (4):where, L is the distance between the two electrodes; A is the effective cross-sectional area of the membrane sample; R is the measured membrane resistance.

2.9. Single cell performance

For the performance test of DMFC, the membrane electrode assemblies (MEAs) were fabricated by using Nafion® 117, recasting Nafion membrane, x%-Ex-SBA-15n and x%-Ex-MSN composite membranes as the electrolytes. The electrodes were purchased from Hephas Energy Co., Ltd. The electrodes about 374 μm in thickness were carbon papers containing microporous layers loaded with 2 mg PtRu per cm² for the anode and 2 mg Pt per cm² for the cathode. The electrodes and composite membrane were hot pressed to form MEAs with 4 cm² active areas at a pressure of 12.5 kg cm⁻² for 90 s at 90 °C. DMFC performance was evaluated in a cell test station at 60 °C by feeding methanol aqueous solution (2, 4, 6, 8 M) at a fixed rate of 40 mL min⁻¹ to the anode and oxygen gas at a flow rate of 0.4 L min⁻¹ to the cathode.

3. Results and discussion

3.1. Characterization of SBA-15n and MSN powders

3.1.1. Powder XRD. The structures of SBA-15n and MSN particles are characterized by X-ray diffraction, and the powder patterns are shown in Fig. 1. Ex-SBA-15n has three peaks at 2θ = 0.84°, 1.42°, and 1.62°, corresponding to the (100), (110) and (200) diffractions of 2D hexagonal P6mm symmetry.³⁰ Ex-MSN has similar 2D hexagonal structure, with the (100), (110) and (200) diffractions appeared at 2θ = 2.05°, 3.55°, and 4.02°.⁷¹ On the other hand, the as-made particles S-SBA-15n and S-MSN shows only a weak (100) peak, attributing to that the pores of S-SBA-15n and S-MSN are still filled with the pore directing agents and the diffraction contrast between the framework and the mesopores is low.^72,73

Fig. 1 XRD patterns of (A) SBA-15n and (B) MSN of (a) extracted, and (b) as-made particles.

3.1.2. N₂ sorption isotherms. The N₂ sorption isotherms of SBA-15n and MSN are shown in Fig. 2. Both ethanol extracted materials display isotherms of type IV (Fig. 2A), which is typical for materials containing ordered mesopores.³⁰ The rapid increases in adsorption volume around P/P₀ = 0.8 for Ex-SBA-15n and P/P₀ = 0.4 for MSN correspond to the capillary condensation of mesopores of 8.1 and 2.6 nm in diameter (Fig. 2B), respectively, while those at P/P₀ > 0.9 are due to the condensation in the voids in between nanoparticles. In contrast, the as-made S-SBA-15n and S-MSN display type II isotherms, which correspond to non-porous materials because the mesopores are blocked by P123 and CTMA⁺. The textural properties derived from N₂ sorption isotherms are shown in Table 1. The surface areas of as-made and extracted SBA-15 are 39 and 514 m² g⁻¹ and the pore volumes are 0.18 and 1.28 cm³ g⁻¹, respectively, while the surface areas of as-made and extracted MSN are 204 and 936 m² g⁻¹ and the pore volumes are 0.20 and 0.95 cm³ g⁻¹, respectively. Since the as-made samples have the surface areas only contributed from the external surfaces of the particles, these results demonstrate that MSN contains higher external surface area than SBA-15n.

Fig. 2 (A) N₂ sorption isotherms (y-axis moved 400 unit apart) and (B) BJH pore size distributions of (a) Ex-SBA-15n (◇), (b) Ex-MSN (△), (c) S-SBA-15n (◆) and (d) S-MSN (▲).

Table 1 Physicochemical properties of Ex-SBA-15n, S-SBA-15n, Ex-MSN and S-MSN mesoporous particles

Sample	d(100) (nm)	a0^a (nm)	SBET (m^2 g^−1)	Vtotal (cm^3 g^−1)	Φp^b (nm)
a a0 = 2d(110)/√3.b Pore diameter, determined from the peak position of BJH distribution curve using the desorption profile of isotherms.
Ex-SBA-15n	10.4	12.0	514	1.3	8.1
S-SBA-15n	10.3	11.9	39	0.18	—
Ex-MSN	4.3	5.0	936	0.95	2.6
S-MSN	4.6	5.3	204	0.20	—

---

3.1.3. SEM images. Fig. 3 shows the SEM images of SBA-15n and MSN materials. The as-made and ethanol extracted materials have similar morphologies and sizes. SBA-15n has rod shape and particle sizes of approximately 100–300 nm, while MSN has spherical shape and diameter of around 100 nm. However, the surfaces of MSN particles are very smooth, while those of SBA-15n contain channels and holes.

Fig. 3 SEM photographs of (a) Ex-SBA-15n, (b) S-SBA-15n, (c) Ex-MSN and (d) S-MSN materials.

3.1.4. Thermal gravimetric analysis for filler materials. Thermal gravimetric analysis (TGA) is used to examine the thermal stabilities of the mesoporous nanoparticles. The TG profile of S-SBA-15n (Fig. 4a) has about 60% weight loss consisting three steps. The first weight loss below 180 °C is due to the loss of absorbed moisture, an abrupt weight loss at 180–200 °C is assigned to the decomposition of the P123 pore directing agent, and the slow weight loss above 200 °C is probably the burning of residual carbon fragments of P123. The weight losses become relatively small for Ex-SBA-15 (Fig. 4b). The weight loss at 180–200 °C is almost disappeared, implying that P123 is removed quite thoroughly. However, there is still ca. 5% weight loss above 200 °C on Ex-SBA-15 and that is attributed to the small amount of residual P123. The TGA profiles of S-MSN and Ex-MSN are shown in Fig. 4c and d. The as-made S-MSN has weight losses in three steps (Fig. 4c). The first step below 200 °C is due to the loss of physically adsorbed water, which is only 4%, inferring the surface is rather hydrophobic. The second step at 200–400 °C is assigned to decomposition of CTMA⁺ surfactant, and the last weight loss above 600 °C is due to the condensation of surface silanol groups.^74,75 Similar high temperature weight loss is also observed on the TGA profile of Ex-MSN, which has the main weight loss of 26% below 200 °C. Other than the low and high temperature weight losses, those between 170 to 530 °C are due to the decomposition and burning of residual CTMA⁺ surfactant. It is noticeable that Ex-MSN has more obviously weight losses below 200 °C and above 600 °C than Ex-SBA-15n, inferring that Ex-MSN contains more surface silanol groups than Ex-SBA-15 and the surface of Ex-MSN is more hydrophilic. On the other hand, the surfaces of as-made mesoporous nanoparticles are more hydrophobic than the ethanol extracted counterparts due to surface coverages with surfactants, which have the hydrophilic parts adsorbed on the silica surfaces and the hydrophobic parts stretch outward. The lower surface hydrophilicity on S-MSN than on S-SBA-15n (Fig. 4c vs. 4a) demonstrate that the long hydrocarbon chain of CTMA⁺ surfactant is more hydrophobic than the poly(propylene glycol) chain of P123. Anyway, the decomposition temperatures of P123 and CTMA⁺ are at 170 and 200 °C, respectively. As a result, the surfactants inside as-made S-SBA-15n and S-MSN can be retained at the operating temperature of DMFCs, which is usually 60–90 °C.

Fig. 4 TGA profiles of (a) S-SBA-15n, (b) Ex-SBA-15n, (c) S-MSN and (d) Ex-MSN materials.

3.1.5. ²⁹Si MAS NMR spectra. The solid state ²⁹Si MAS NMR spectra of Ex-SBA-15n and Ex-MSN are showed in Fig. 5. Three distinct resonance peaks corresponding to Q_n (Q_n = Si(OSi)_n(OH)_4−n), where n = 2–4 are observed at −111 ppm for Q₄, −101 ppm for Q₃, and −92 ppm for Q₂.⁷⁶ For Ex-SBA-15n, the most intense peak is Q₄. However, the most intense peak for Ex-MSN is Q₃. These results confirm that Ex-MSN has more silanol groups on surface than Ex-SBA-15n. The larger amount of silanol groups on Ex-MSN makes this material more hydrophilic than Ex-SBA-15, as demonstrated in the TGA results.

Fig. 5 Solid state ²⁹Si MAS NMR spectra of (a) Ex-SBA-15n and (b) Ex-MSN materials.

3.2. Characterization of composite membranes

3.2.1. Mapping images. The dot-mapping images of silicon in the cross-sections of mesoporous silica nanoparticles/Nafion composite membranes with various loadings of SBA-15n and MSN are displayed in Fig. S3–S6 (ESI†). The images show that Si is well dispersed in the Nafion matrix up to the loading of 20 wt% SBA-15n and 10 wt% MSN. The highest loading of Ex-MSN is only 10 wt% loading, because the cell performance assembled from Nafion matrix with higher than 10 wt% loading of Ex-MSN becomes worse. The good dispersion and relatively high loading amounts of SBA-15n and MSN in the composite membrane are due to two reasons. One is that mesoporous silica materials are kept wet in the synthesis procedure to avoid aggregation of nanoparticles. The other is the mesoporous silica materials are in nanosize.

3.2.2. Dynamic mechanical analysis (DMA). The mechanical properties of the membranes were evaluated by dynamic mechanical analysis in tension mode at 35 °C. The results are presented in Fig. S7 (ESI†) and the calculated Young's moduli are summarized in Table 2. The Young's moduli for the recasting Nafion membrane and Nafion® 117 membranes are 46 and 174 MPa, respectively. The more than triple strength of commercial Nafion® 117 membrane than the recasting one implies that the commercial membrane is denser and mechanically stronger due to different preparation processes. The moduli of composite membranes filled with SBA-15 and MSN are all higher than that of recast Nafion membrane, and the values increase with filler amounts. The data show that the incorporation of nanoparticles of mesoporous materials can effectively improve the mechanical strength of recast membranes. Moreover, the Young's moduli are similar for the composites with the same amounts of fillers with or without the surfactants in the pores. Therefore, the enhancement in the Young's moduli of composite membranes is probably due to the strong electrostatic attractions at Nafion/SBA-15n or MSN interface, which efficiently restricts the movement of the polymer chains of Nafion matrix and thus make the stretching difficult. It is also noticeable that the moduli of Nafion composite membranes filled with SBA-15n are much higher than those filled with MSN, implying that stronger interaction is present in Nafion/SBA-15n interfaces than in Nafion/MSN interfaces.

Table 2 Young's moduli of Nafion® 117, recasting Nafion membrane, Ex-SBA-15n, S-SBA-15n, Ex-MSN and S-MSN mesoporous particles

Membrane	Young's moduli (MPa)	Membrane	Young's moduli (MPa)
N117	174	Recasting	46
1%-Ex-SBA-15n	149	1%-S-SBA-15n	178
5%-Ex-SBA-15n	190	5%-S-SBA-15n	234
10%-Ex-SBA-15n	254	10%-S-SBA-15n	304
15%-Ex-SBA-15n	334	15%-S-SBA-15n	325
20%-Ex-SBA-15n	424	20%-S-SBA-15n	435
1%-Ex-MSN	67	1%-S-MSN	62
2.5%-Ex-MSN	70	2.5%-S-MSN	66
5%-Ex-MSN	71	5%-S-MSN	74
10%-Ex-MSN	95	10%-S-MSN	76

---

3.2.3. TGA for membranes. The thermal stabilities of the membranes were analysed by taking TGA. As shown in Fig. S8 (ESI†), all the samples show similar weight losses profiles. The first step of weight losses occurs at 100–200 °C, assigning to the evaporation of residual solvents and moisture. The second step weight loss at about 350 °C is assigned to the degradation of sulfonic acid groups on the framework of Nafion polymers.⁷⁷ The third step weight loss occurs at about 500 °C, corresponding to the degradation of polymer main chains.⁷⁷ The observation that almost no residues present above 500 °C is probably due to that the silica fillers in nano-sizes were blew away by the carrier gas steam after polymer decomposition. Nafion® 117 possessed slightly higher thermal stability than other membranes, again due to different preparation processes. Nevertheless, thermal stabilities of all recasting membranes are similar and the structures can be retained up to 350 °C.

3.2.4. SAXS. The structural properties of recasting Nafion membrane and that filled with 5% S-SBA-15n composite membrane were characterized by SAXS. The results are shown in Fig. 6. For dry recasting Nafion membrane (Fig. 6a), one broad peak with the maximum at d-spacing of 3.3 nm is seen. The peak maximum shifts towards lower-q after the membrane is soaked in water for 24 h (Fig. 6b), implying that water is soaked into the ionic domains of Nafion membrane and the water channels are formed by swelling the ionic domains. The SAXS pattern of the swollen recasting membrane shows a strong broad peak with the maximum at d-spacing of 4.8 nm, implying good ordering of the water channels. In comparison, Fig. 6c shows that the dry 5%-S-SBA-15n composite membrane reveals similar broad SAXS peak at d = 3.3 nm, in addition to a sharp peak with d-spacing of 11 nm, which is close to the d₍₁₀₀₎ spacing of SBA-15 materials observed in XRD analysis. The additional peak probably comes from the ordered pore structure of SBA-15. After soaking the 5%-S-SBA-15n composite membrane in water (Fig. 6d), the sharp peak is almost invisible, presumably due to low scattering contrast between the silica framework of SBA-15 and the water filled pores. Besides, the swollen composite membrane contains water channels of similar size as that in the pristine Nafion membrane. Since the particle size of SBA-15 filler is larger than the diameter of water channels, and the water channels are not significantly expanded with SBA-15 fillers, the SBA-15 filler should probably locate around the water channels instead of inside the channels.

Fig. 6 SAXS patterns for the (a) dry and (b) wet recasting Nafion membrane and (c) dry and (d) wet 5%-S-SBA-15n composite membrane.

3.2.5. Methanol permeability. The methanol permeabilities of the membranes are measured at 35 °C, and the changes of methanol permeability as a function of SBA-15n and MSN loadings are shown in Fig. 7. These data are compared with two reference Nafion membranes. One is the commercial Nafion® 117 membrane, and the other is the recasting membrane. The commercial one has lower methanol permeability than the recasting one, due to different preparation processes. All composite membranes have lower methanol permeability than recasting Nafion membrane, indicating that the SBA-15n and MSN particles can effectively block the water channels.⁵⁷ Fig. 7A compares the effect of SBA-15n with and without P123 pore-directing agent as the filler. The methanol permeability is found to decrease almost linearly with the increase of S-SBA-15n loading, while less influence is observed for Ex-SBA-15n. This result indicates that P123 present in the mesopores of SBA-15n can effectively resist methanol transfer through the pores.

Fig. 7 Methanol permeability at 35 °C of Nafion composite membranes with different loadings of (A) Ex-SBA-15n () vs. S-SBA-15n (), (B) Ex-MSN () vs. S-MSN (), (C) S-SBA-15n () vs. S-MSN () and (D) Ex-SBA-15n () vs. Ex-MSN () in comparison to those of Nafion® 117 () and recasting Nafion membranes ().

Fig. 7B compares the effect of MSN with and without CTMA⁺ surfactant as the filler. Lower methanol permeabilities are obtained over composite membranes containing S-MSN than those with Ex-MSN, again implying the surfactant inside the mesopores can resist methanol transfer through the pores. Methanol permeability also decreases with the increase of MSN loading until ca. 5 wt%. Further increase of MSN loading has little influence or even increases methanol permeability. This can be ascribed to the segregation of inorganic fillers and microcracking creating diffusion paths at the MSN/Nafion interface.^78,79

Fig. 7C compares S-SBA-15n and S-MSN fillers both containing the pore-directing agents in the mesopores. With fillers loadings up to 10 wt%, the methanol permeabilities of the composite membranes containing S-MSN are lower than those of S-SBA-15n. This result is probably due to that CTMA⁺ is more hydrophobic than P123. It is also noticeable that all the composite membranes with 1–10 wt% S-MSN fillers and those with >5 wt% S-SBA-15n fillers have methanol permeabilities lower than commercial Nafion® 117 membrane. As to the mesoporous nanoparticles without the pore-directing agents, Ex-MSN and Ex-SBA-15n, Fig. 7D shows that the methanol permeabilities of the composite membranes do not obviously change with the loading. SBA-15n and MSN have pore diameter of ca. 8 nm and 2.6 nm, respectively, which are much larger than the size of methanol molecule (0.41 nm). As a result, methanol may transfer through the mesopores of Ex-SBA-15n and Ex-MSN. However, the methanol permeabilities of the composite membranes containing Ex-MSN and Ex-SBA-15n still lower than the recasting Nafion membrane and close to the commercial Nafion® 117 membrane.

3.2.6. Proton conductivity. The proton conductivities of SBA-15n and MSN composite membranes with different nanoparticle contents are measured at 90 °C, and the results are compared with those of recasting Nafion membrane and commercial Nafion® 117 membrane, as shown in Fig. 8 and Table 3. Commercial Nafion® 117 membrane has much higher proton conductivity than the recasting membrane, attributing to different preparation methods. Fig. 8A compares the effect of SBA-15n with and without P123 pore-directing agent as the filler. A maximum proton conductivity is obtained on the composite membrane containing 5% filler for both S-SBA-15n and Ex-SBA-15n. It is also noticed that the proton conductivities of composite membranes containing S-SBA-15n are higher than those containing Ex-SBA-15n with the same loading of fillers, attributed to that the ether groups on pore-directing agent P123 may assist proton transfer through oxygen atoms. All composite membranes, except 20%-Ex-SBA-15n, give higher proton conductivities than recasting Nafion membrane. Among them, only 5%-S-SBA-15n has higher proton conductivities than commercial Nafion membrane. The enhanced proton conductivities of SBA-15n/Nafion membranes are probably due to the formation of new ionic channels at interphase region of the SBA-15 nanoparticles and Nafion matrix.^79,80 That is proved by the SAXS results that the SBA-15 fillers probably locate around the water channels instead of inside the channels. However, too high loading of silica nanoparticles may occlude the hydrophilic clusters of the Nafion membrane and block proton transfer.^81,82

Fig. 8 Proton conductivities at 90 °C and water uptakes at ambient temperature of Nafion composite membranes with different loadings of (A) Ex-SBA-15n vs. S-SBA-15n, (B) Ex-MSN vs. S-MSN, (C) S-SBA-15n vs. S-MSN and (D) Ex-SBA-15n vs. Ex-MSN in comparison to those of Nafion® 117 and recasting Nafion membranes.

Table 3 Characterization and single cell performances of Nafion® 117, recasting Nafion membrane and composite membranes

Membrane	σ^a (10^−2 S cm^−1)	P^b (10^−6 cm s^−1)	Φ^c (10^3 S cm^−3 s)	WU^d (wt%)	OCV (V)	PD^e (mW cm^−2)
a Proton conductivity tested at 90 °C and 90% percentage humidity.b Methanol permeability obtained from the methanol concentration of diffusion reservoir at 35 °C.c Selectivity, Φ = σ/P.d Water uptake tested at room temperature, WU = Wwet − Wdry/Wdry.e Power density test with 2 M MeOH at 60 °C.
Nafion® 117	3.39	3.87	8.76	25	0.76	96
Recasting	2.56	4.21	6.08	38	0.73	65
1%-Ex-SBA-15n	2.62	3.70	7.08	37	0.73	41
2.5%-Ex-SBA-15n	3.02	3.74	8.07	36	0.73	79
5%-Ex-SBA-15n	3.38	3.72	9.09	45	0.80	109
10%-Ex-SBA-15n	3.12	3.78	8.25	36	0.73	68
15%-Ex-SBA-15n	2.85	3.66	7.79	32	—	—
20%-Ex-SBA-15n	2.49	3.58	6.96	30	—	—
1%-S-SBA-15n	2.81	4.19	6.71	32	0.78	93
2.5%-S-SBA-15n	3.39	4.07	8.33	43	0.78	104
5%-S-SBA-15n	3.94	3.84	10.26	55	0.79	117
10%-S-SBA-15n	3.45	3.48	9.91	49	0.80	107
15%-S-SBA-15n	3.38	3.20	10.56	37	0.72	64
20%-S-SBA-15n	3.01	2.71	11.11	33	—	—
1%-Ex-MSN	3.46	4.01	8.63	44	0.76	98
2.5%-Ex-MSN	3.76	3.89	9.67	46	0.77	122
5%-Ex-MSN	4.01	3.50	11.46	59	0.76	131
10%-Ex-MSN	3.44	3.88	8.87	50	0.78	107
1%-S-MSN	3.06	3.45	8.87	40	0.77	116
2.5%-S-MSN	2.90	2.99	9.70	43	0.73	107
5%-S-MSN	2.70	2.89	9.34	55	0.70	97
10%-S-MSN	2.55	3.02	8.44	59	0.74	69

---

Fig. 8B compares the effect of MSN with and without CTMA⁺ surfactant as the filler. The proton conductivities of composite membranes containing S-MSN are all lower than those containing Ex-MSN, with the same loading of fillers. It is attributed to that the CTMA⁺ surfactant in the mesopores of S-MSN has its positive charged head strongly interacted with the silicate (Si–O⁻) wall, while the hydrocarbon chain obstructs the proton transfer through the mesopores. Among the composite membranes containing Ex-MSN fillers, those of 2.5% and 5% give higher proton conductivities than commercial Nafion® 117 membrane. If comparing with the recasting Nafion membrane, all composite membranes with MSN fillers, except 10%-S-MSN, give higher proton conductivities.

The proton conductivity of S-SBA-15n and S-MSN composite membranes are compared and shown in Fig. 8C. Except for 1%-S-SBA-15n, the proton conductivities of composite membranes containing S-SBA-15n are higher than those containing S-MSN at same filler loading. That is due to that ether groups of P123 in S-SBA-15n may assist proton transfer, while the CTMA⁺ obstructs the proton transfer through the mesopores of S-MSN. As to the slightly higher proton conductivity of 1% S-MSN than 1% S-SBA-15n, it is probably owing to higher surface area of S-MSN than S-SBA-15n (Table 1) and more ionic channels are formed at interphase region of S-MSN and Nafion matrix.

Fig. 8D compares Ex-SBA-15n and Ex-MSN fillers both without the pore-directing agents in the mesopores, and shows that the proton conductivities of all composite membranes containing Ex-MSN are higher than those containing Ex-SBA-15n. That is attributed to that the Ex-MSN materials have higher surface areas and more silanol groups on the surfaces than Ex-SBA-15n, based on the results from N₂ sorption (Table 1), TGA (Fig. 4a and c), and ²⁹Si MAS NMR spectra (Fig. 5). The silanol groups inside the mesopores provide additional hopping sites and migration route, while the outer surface areas of nanoparticles enhance interphase region between mesoporous silica materials and Nafion matrix for proton conductivity. An optimum proton conductivity of 4.01 × 10⁻² S cm⁻¹ is obtained on 5%-Ex-MSN/Nafion composite membrane. These results indicate that the silanol groups on surfaces of mesoporous materials can effectively enhance proton conductivity.

The activation energies of proton conductivities in different membranes were estimated from the Arrhenius plots (Fig. S9, ESI†). The activation energies are similar, 7.88 and 7.47 kJ mol⁻¹, respectively, for Nafion® 117 and recasting Nafion membranes. The activation energies are 9.44, 5.92, 5.42 and 11.17 kJ mol⁻¹ for Nafion membranes filled with Ex-SBA-15n, S-SBA-15n, Ex-MSN and S-MSN, respectively. The lowest activation energies are obtained on composite membranes filled with S-SBA-15n and Ex-MSN, and these results are in consistency with the trend in proton conductivities. It is generally considered that the proton transport in the membrane follows Grotthuss and vehicle mechanisms. For the Grotthuss mechanism, the activation energy for proton conduction was reported to be around 14–40 kJ mol⁻¹.^83–85 From the activation energies measured in our composite membranes, vehicle mechanism should be the main pathway responsible for the proton conductivity.

3.2.7. Water uptake. Water uptake (WU) is an important property of proton exchange membrane (PEM). One of the proton conduction mechanisms in PEM is vehicular mechanism, where the water solvated protons (H⁺(H₂O)_x) transfer through the membrane under the electro-osmotic drag.^9,77 Therefore, the proton conductivity of the membrane is usually increases with WU of the membrane. The WU values of the composite membranes as a function of filler loading are also shown in Fig. 8. Fig. 8A and D demonstrate that the WU values are almost parallel to the proton conductivities, implying that vehicular mechanism plays an important role in proton conductivities in these membranes. In contrast, the composite membranes containing S-MSN shown in Fig. 8B and C give the WU values varying in a reverse trend from that of proton conductivities. The WU increases while the proton conductivity decreases with the filler loading. TG analysis shows that the surfaces of S-MSN nanoparticles are very hydrophobic, due to the coverage of particle surfaces with the CTMA⁺ surfactant. However, the surface CTMA⁺ surfactant should be removed and washed away during membrane treatment with hot aqueous solutions of sulfuric acid and H₂O₂. The increase in WUs of composite membranes with the loading of S-MSN nanoparticles is likely due to the increase in hydrophilic silanol groups in the Nafion matrix. Nevertheless, the S-MSN nanoparticles are probably imbedded in the water channels of Nafion matrix and hinder proton transfer.

3.2.8. Selectivity. The selectivity of the membrane is calculated by the ratio of proton conductivity over methanol permeability, and it is generally used as a comprehensive index to evaluate the performance of DMFC membranes.⁸⁶ The selectivities of the composite membranes as a function of filler loading are shown in Fig. 9. Again, commercial Nafion membrane has higher selectivity than recasting one due to different preparation process. Fig. 9A compares the effect of SBA-15n with and without P123 pore-directing agent as the filler. The selectivities of S-SBA-15n composite membranes are higher than those of Ex-SBA-15n composite membranes, implying that the presence of P123 pore-directing agent in the pores of SBA-15 nanoparticles is favorable. For S-SBA-15n composite membranes, a rapid increase in selectivity up to 5 wt% loading is seen, and then it slows down. Moreover, these selectivities are higher than that of commercial Nafion membrane. On the other hand, Ex-SBA-15n composite membranes have the maximum selectivity at 5 wt% loading, and the value is just slightly higher than that of commercial Nafion membrane. Fig. 9B compares the effect of MSN with and without CTMA⁺ surfactant as the filler. It is noticeable that most of the composite membranes filled with MSN have higher selectivities than commercial Nafion membrane. The maximum selectivity is obtained on the composite membrane containing 5 wt% Ex-MSN. Fig. 9C compares the selectivities of composite membranes containing the nanoparticle fillers with pore-directing agents retained in the mesopores. Highest selectivities are observed on the composite membranes containing 5–20% S-SBA-15n. Fig. 9D compares the selectivities of composite membranes containing the nanoparticle fillers without pore-directing agents. The selectivities of Ex-MSN composite membranes are all higher than those of Ex-SBA-15n composite membranes, implying that the presence of large amount of silanol groups on the surfaces of nanoparticles is favorable.

Fig. 9 Selectivities of Nafion composite membranes with different loadings of (A) Ex-SBA-15n () vs. S-SBA-15n (), (B) Ex-MSN () vs. S-MSN (), (C) S-SBA-15n () vs. S-MSN () and (D) Ex-SBA-15n () vs. Ex-MSN () in comparison to those of Nafion® 117 () and recasting Nafion membranes ().

3.2.9. Cell performance. The performances of the MEAs fabricated by using composite membranes as the electrolytes in DMFC are compared with those assembled with Nafion® 117 and recasting Nafion membranes under the operation condition of 2, 4, 6 and 8 M methanol solution at 60 °C. Fig. 10 and 11 show that all MEAs have power densities (PDs) decreased with the increase of methanol concentration. It is attributed to that the methanol crossover becomes faster at higher methanol concentration, and that reduces the cell voltage as well as power density. Fig. 9 compares the MEAs assembled with composite membranes containing SBA-15n with (Fig. 10A) and without (Fig. 10B) P123 pore-directing agent. Generally, the MEAs with S-SBA-15n possess higher power densities than those with Ex-SBA-15n composite membranes of the same loading, in consistency with the variation in selectivities of the membranes. Among the MEAs assembled with SBA-15n composite membranes, those with 1–10% S-SBA-15n all have higher PDs than the MEA with recasting Nafion membrane at different methanol concentrations. Moreover, the MEAs giving PDs higher than that of MEA with commercial N117 membrane are assembled with 5%-Ex-SBA-15n and 2.5–10% S-SBA-15n composite membranes. These results infer that S-SBA-15n which have P123 surfactant inside the mesopores can effectively enhance the PDs of the MEAs by improving proton transfer and hindering methanol permeability.

Fig. 10 Power density of DMFC single cells assembled with Nafion composite membranes with different loadings of (A) S-SBA-15n and (B) Ex-SBA-15n in comparison to those of Nafion® 117 () and recasting Nafion membranes () under the operation condition of 2, 4, 6 and 8 M methanol at 60 °C (, : 1%, , : 2.5%, , : 5%, , : 10%, : 15%).

Fig. 11 Power density of DMFC single cells assembled with Nafion composite membranes with different loadings of (A) S-MSN and (B) Ex-MSN in comparison to those of Nafion® 117 () and recasting Nafion membranes () under the operation condition of 2, 4, 6 and 8 M methanol at 60 °C (, : 1%, , : 2.5%, , : 5%, , : 10%).

Fig. 11 compares the MEAs assembled with composite membranes containing MSN with (Fig. 11A) and without (Fig. 11B) CTMA⁺ surfactant. The MEAs containing Ex-MSN composite membranes possess higher power density than those with S-MSN composite membranes of the same nanoparticle loading. These results are consistent with the proton conductivity measurement, which shows that proton transfer is obstructed by CTMA⁺ inside the mesopores of S-MSN. The PDs of all the MEAs assembled with Ex-MSN composite membrane are higher than that with recasting membrane at all methanol concentrations, and the PDs of MEAs assembled with 2.5 and 5% Ex-MSN are even higher than that with commercial N117 membrane. According to Fig. 9 and 10, the pore-directing agents P123 and CTMA⁺ inside the mesopores of silica nanoparticles influence the power densities of MEAs very differently.

The IV curves of MEAs assembled with different membranes using 2 M methanol solution as the fuel at 60 °C are shown in Fig. 12 and 13 and the related data are tabulated in Table 3. The open circuit voltages are similar over all MEAs assembled with the composite membranes, attributing to that the methanol permeabilities do not change dramatically with different loadings of inorganic fillers. Nevertheless, the membranes with lower methanol permeabilities can retain at higher voltages with the increase in current density. Fig. 12 shows that the MEAs assembled with S-SBA-15n composite membranes possess higher power density than those with Ex-SBA-15n composite membranes. Moreover, the membranes with 5 wt% loading of both Ex-SBA-15n and S-SBA-15n fillers give the highest power densities among different loadings. The MEA assembled with 5 wt% S-SBA-15n composite membrane has the highest power density of 117 mW cm⁻², which is higher than the MEA with recasting Nafion membrane of 65 mW cm⁻² and commercial Nafion® 117 membrane of 96 mW cm⁻². The trend is different for MEAs assembled with MSN composite membranes (Fig. 13). The MEAs assembled with S-MSN composite membranes possess lower power densities than those with Ex-MSN composite membranes. Moreover, the MEAs assembled with 5%-Ex-MSN composite membrane possesses the highest power density of 131 mW cm⁻², which is higher than that assembled with 5%-S-SBA-15n composite membrane of 117 mW cm⁻². The results indicate that the amount of silanol groups on the surfaces of mesoporous fillers is the most important influence on proton transfer and power density of MEA.

Fig. 12 Comparison of DMFC performances at 2 M methanol feed concentration and 60 °C by using membranes of (A) x%-S-SBA-15n and (B) x%-Ex-SBA-15 with different loadings in comparison to those with Nafion® 117 and recasting Nafion membranes.

Fig. 13 Comparison of DMFC performances at 2 M methanol feed concentration and 60 °C by using membranes of (A) x%-S-MSN and (B) x%-Ex-MSN with different loadings in comparison to those with Nafion® 117 and recasting membranes.

3.2.10. Long-term test. In order to verify the practical usage of the novel composite membranes developed in this study, the cells assembled with composite membranes filled with 5%-S-SBA-15n and 5%-Ex-MSN, which possess the highest power densities are examined for long-term test, and the results are compared with that assembled with recasting Nafion membrane. The OCVs of DMFC were recorded at 60 °C for 5500 min using 2 M methanol and 0.4 mL min⁻¹ oxygen on the anode and cathode, respectively. As shown in Fig. 14, there is an initial and slight drop in OCV in the first 50 min for all these three MEAs. Then, the OCVs are stabilized and the drop is negligible, suggesting no degradation of the membranes up to 5500 min. Moreover, the cell assembled with composite membranes filled with 5%-S-SBA-15n gives higher OCV than the other two cells.

Fig. 14 Long-term tests of 5%-S-SBA-15n (upper, red line), 5%-Ex-MSN (lower, green line) composite membrane and recasting Nafion membrane (lower, black line) at OCV under fuel cell configuration.

4. Conclusions

In this work, the nanoparticles of mesoporous silica (SBA-15n and MSN) were successfully prepared and incorporated in Nafion to form composite membranes by recasting technique. The performances of the membranes were compared in terms of the loadings of SBA-15n and MSN, the influence of different pore-directing agents, and the presence or removal of pore-directing agents inside the mesopores of silica fillers. The results indicated that the pore-directing agents (P123 and CTMA⁺) in nanoparticles of mesoporous silica could effectively resist the methanol penetration from anode to cathode. Moreover, the methanol permeability decreased almost linearly with the increase of S-SBA-15n loading in composite membrane and the lowest methanol permeability of 2.71 × 10⁻⁶ cm s⁻¹ was observed on composite membrane containing 20% S-SBA-15n. S-MSN also had contribution of blocking methanol permeability, but the loadings above 2.5% had similar effect of ca. 3.0 × 10⁻⁶ cm s⁻¹ permeability. In contrast, ethanol extracted nanoparticles, both Ex-SBA-15n and Ex-MSN showed little contribution in inhibiting methanol permeability due to too large pore diameters in comparison to the size of methanol. All composite membranes show higher proton conductivities than the recasting Nafion membrane at 90 °C, indicating that new ionic channels might be created in the interphase region of fillers/polymer for proton transfer. Moreover, P123 in S-SBA-15n provide additional proton transfer routes inside the mesopores through the oxygen atoms on the block copolymer of ethylene glycol and propylene glycol. In contrast, CTMA⁺ in S-MSN obstructs the proton transfer through the mesopores. For the composite membranes filled with extracted mesoporous nanoparticles, the silanol groups on the surfaces of empty mesopores would assist proton transfer. As a result, Ex-MSN which contained a larger amount of silanol groups than Ex-SBA-15n contributed to higher proton conductivities of Ex-MSN composite membranes than Ex-SBA-15n composite membranes. Except for S-MSN, there were an optimal loading of ca. 5% for Ex-MSN, Ex-SBA-15n and S-SBA-15n to give highest proton conductivities. However, the highest selectivities were obtained on the composite membranes filled with 5%-Ex-MSN and 20%-S-SBA-15n. Among all composite membranes, that filled with 5%-Ex-MSN showed the highest proton conductivity of 4.01 × 10⁻² cm s⁻¹ and the single cell assembled with 5%-Ex-MSN membrane also gave the highest power density of 131 mW cm⁻², which was about 2 times higher than the cell with recasting Nafion membrane and 36% higher than that with commercial Nafion® 117 membrane.

Acknowledgements

Financial supports from Ministry of Science & Technology and Ministry of Education, Taiwan are gratefully acknowledged. Acknowledgments are also extended to Ms Su-Jen Ji and Chia-Ying Chien of the Instrumentation Center, National Taiwan University for SEM and TEM experiments. The assistances of Dr U-Ser Jeng and Ms Kuei-Fen Liao of the NSRRC, Taiwan are gratefully acknowledged for the SAXS experiments.

References

1. V. Neburchilov, J. Martin, H. J. Wang and J. J. Zhang, J. Power Sources, 2007, 169, 221–238.
2. X. M. Ren, M. S. Wilson and S. Gottesfeld, J. Electrochem. Soc., 1996, 143, L12–L15.
3. S. Surampudi, S. R. Narayanan, E. Vamos, H. Frank, G. Halpert, A. Laconti, J. Kosek, G. K. S. Prakash and G. A. Olah, J. Power Sources, 1994, 47, 377–385.
4. K. Scott, W. M. Taama, P. Argyropoulos and K. Sundmacher, J. Power Sources, 1999, 83, 204–216.
5. K. D. Kreuer, J. Membr. Sci., 2001, 185, 29–39.
6. M. P. Hogarth and T. R. Ralph, Platinum Met. Rev., 2002, 46, 146–164.
7. M. T. Wang, K. W. Feindel, S. H. Bergens and R. E. Wasylishen, J. Power Sources, 2010, 195, 7316–7322.
8. J. Jagur-Grodzinski, Polym. Adv. Technol., 2007, 18, 785–799.
9. S. J. Peighambardoust, S. Rowshanzamir and M. Amjadi, Int. J. Hydrogen Energy, 2010, 35, 9349–9384.
10. P. L. Antonucci, A. S. Arico, P. Creti, E. Ramunni and V. Antonucci, Solid State Ionics, 1999, 125, 431–437.
11. A. K. Sahu, G. Selvarani, S. Pitchumani, P. Sridhar and A. K. Shukla, J. Appl. Electrochem., 2007, 37, 913–919.
12. G. Lakshminarayana and M. Nogami, Electrochim. Acta, 2010, 55, 1160–1168.
13. A. K. Sahu, S. Pitchumani, P. Sridhar and A. K. Shukla, Bull. Mater. Sci., 2009, 32, 285–294.
14. A. K. Sahu, A. Jalajakshi, S. Pitchumani, P. Sridhar and A. K. Shukla, J. Chem. Sci., 2012, 124, 529–536.
15. A. K. Sahu, G. Selvarani, S. Pitchumani, P. Sridhar and A. K. Shukla, J. Electrochem. Soc., 2007, 154, B123–B132.
16. M. Watanabe, H. Uchida and M. Emori, J. Phys. Chem. B, 1998, 102, 3129–3137.
17. Y. Patil, S. Sambandam, V. Ramani and K. Mauritz, J. Electrochem. Soc., 2009, 156, B1092–B1098.
18. D. A. Siuzdak, P. R. Start and K. A. Mauritz, J. Polym. Sci., Part B: Polym. Phys., 2003, 41, 11–22.
19. S. Siracusano, V. Baglio, M. A. Navarra, S. Panero, V. Antonucci and A. S. Arico, Int. J. Electrochem. Sci., 2012, 7, 1532–1542.
20. J. J. Pan, H. N. Zhang, W. Chen and M. Pan, Int. J. Hydrogen Energy, 2010, 35, 2796–2801.
21. A. K. Shukla, A. K. Sahu, S. Pitchumani and P. Sridhar, Fuel Cells, 2009, 9, 139–147.
22. V. Tricoli and F. Nannetti, Electrochim. Acta, 2003, 48, 2625–2633.
23. P. Dimitrova, K. A. Friedrich, U. Stimming and B. Vogt, Solid State Ionics, 2002, 150, 115–122.
24. Z. G. Shao, P. Joghee and I. M. Hsing, J. Membr. Sci., 2004, 229, 43–51.
25. B. Tazi and O. Savadogo, Electrochim. Acta, 2000, 45, 4329–4339.
26. E. Abouzari-Iotf, M. M. Nasef, H. Ghassemi, M. Zakeri, A. Ahmad and Y. Abdollahi, ACS Appl. Mater. Interfaces, 2015, 7, 17008–17015.
27. D. H. Jung, S. Y. Cho, D. H. Peck, D. R. Shin and J. S. Kim, J. Power Sources, 2003, 118, 205–211.
28. P. Costamagna, C. Yang, A. B. Bocarsly and S. Srinivasan, Electrochim. Acta, 2002, 47, 1023–1033.
29. G. Alberti, M. Casciola, U. Costantino and M. Leonardi, Solid State Ionics, 1984, 14, 289–295.
30. D. Y. Zhao, J. L. Feng, Q. S. Huo, N. Melosh, G. H. Fredrickson, B. F. Chmelka and G. D. Stucky, Science, 1998, 279, 548–552.
31. J. S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, K. D. Schmitt, C. T. W. Chu, D. H. Olson, E. W. Sheppard, S. B. Mccullen, J. B. Higgins and J. L. Schlenker, J. Am. Chem. Soc., 1992, 114, 10834–10843.
32. I. I. Slowing, J. L. Vivero-Escoto, B. G. Trewyn and V. S. Y. Lin, J. Mater. Chem., 2010, 20, 7924–7937.
33. Y. Daiko, T. Kasuga and M. Nogami, Microporous Mesoporous Mater., 2004, 69, 149–155.
34. R. Marschall, I. Bannat, J. Caro and M. Wark, Microporous Mesoporous Mater., 2007, 99, 190–196.
35. B. Y. Jiang, H. L. Tang and M. Pan, Int. J. Hydrogen Energy, 2012, 37, 4612–4618.
36. Y. G. Jin, S. Z. Qiao, L. Zhang, Z. P. Xu, S. Smart, J. C. D. da Costa and G. Q. Lu, J. Power Sources, 2008, 185, 664–669.
37. Y. S. Ye, G. W. Liang, B. H. Chen, W. C. Shen, C. Y. Tseng, M. Y. Cheng, J. Rick, Y. J. Huang, F. C. Chang and B. J. Hwang, J. Power Sources, 2011, 196, 5408–5415.
38. A. K. Sahu, S. D. Bhat, S. Pitchumani, P. Sridhar, V. Vimalan, C. George, N. Chandrakumar and A. K. Shukla, J. Membr. Sci., 2009, 345, 305–314.
39. W. H. J. Hogarth, J. C. D. da Costa, J. Drennan and G. Q. Lu, J. Mater. Chem., 2005, 15, 754–758.
40. F. Pereira, K. Valle, P. Belleville, A. Morin, S. Lambert and C. Sanchez, Chem. Mater., 2008, 20, 1710–1718.
41. C. S. Karthikeyan, S. P. Nunes, L. A. S. A. Prado, M. L. Ponce, H. Silva, B. Ruffmann and K. Schulte, J. Membr. Sci., 2005, 254, 139–146.
42. Y. Tominaga, I. C. Hong, S. Asai and M. Sumita, J. Power Sources, 2007, 171, 530–534.
43. Y. F. Lin, C. Y. Yen, C. C. M. Ma, S. H. Liao, C. H. Lee, Y. H. Hsiao and H. P. Lin, J. Power Sources, 2007, 171, 388–395.
44. D. Liu, L. Geng, Y. Q. Fu, X. Dai and C. L. Lu, J. Membr. Sci., 2011, 366, 251–257.
45. J. Wu, Z. M. Cui, C. J. Zhao, H. T. Li, Y. Zhang, T. Z. Fu, H. Na and W. Xing, Int. J. Hydrogen Energy, 2009, 34, 6740–6748.
46. S. J. Park, D. H. Lee and Y. S. Kang, J. Membr. Sci., 2010, 357, 1–5.
47. J. H. Won, H. J. Lee, K. S. Yoon, Y. T. Hong and S. Y. Lee, Int. J. Hydrogen Energy, 2012, 37, 9202–9211.
48. I. S. Amiinu, W. Li, G. J. Wang, Z. K. Tu, H. L. Tang, M. Pan and H. N. Zhang, Electrochim. Acta, 2015, 160, 185–194.
49. L. Xie, E. B. Cho and D. Kim, Solid State Ionics, 2011, 203, 1–8.
50. L. Y. Hong, S. Y. Oh, A. Matsuda, C. S. Lee and D. P. Kim, Electrochim. Acta, 2011, 56, 3108–3114.
51. E. Vijayakumar and D. Sangeetha, RSC Adv., 2015, 5, 42828–42835.
52. S. Angioni, D. C. Villa, A. S. Cattaneo, P. Mustarelli and E. Quartarone, J. Power Sources, 2015, 294, 347–353.
53. Q. Cai, Z. S. Luo, W. Q. Pang, Y. W. Fan, X. H. Chen and F. Z. Cui, Chem. Mater., 2001, 13, 258–263.
54. V. S. Y. Lin, D. R. Radu, M. K. Han, W. H. Deng, S. Kuroki, B. H. Shanks and M. Pruski, J. Am. Chem. Soc., 2002, 124, 9040–9041.
55. R. I. Nooney, D. Thirunavukkarasu, Y. M. Chen, R. Josephs and A. E. Ostafin, Chem. Mater., 2002, 14, 4721–4728.
56. S. Huh, J. W. Wiench, J. C. Yoo, M. Pruski and V. S. Y. Lin, Chem. Mater., 2003, 15, 4247–4256.
57. S. C. Byun, Y. J. Jeong, J. W. Park, S. D. Kim, H. Y. Ha and W. J. Kim, Solid State Ionics, 2006, 177, 3233–3243.
58. J. Lu, M. Liong, Z. X. Li, J. I. Zink and F. Tamanoi, Small, 2010, 6, 1794–1805.
59. I. I. Slowing, B. G. Trewyn, S. Giri and V. S. Y. Lin, Adv. Funct. Mater., 2007, 17, 1225–1236.
60. C. P. Tsai, Y. Hung, Y. H. Chou, D. M. Huang, J. K. Hsiao, C. Chang, Y. C. Chen and C. Y. Mou, Small, 2008, 4, 186–191.
61. S. H. Wu, Y. S. Lin, Y. Hung, Y. H. Chou, Y. H. Hsu, C. Chang and C. Y. Mou, ChemBioChem, 2008, 9, 53–57.
62. C. C. Liu, T. S. Lin, S. I. Chan and C. Y. Mou, J. Catal., 2015, 322, 139–151.
63. K. C. Kao, T. S. Lin and C. Y. Mou, J. Phys. Chem. C, 2014, 118, 6734–6743.
64. F. J. Fernandez-Carretero, V. Compan and E. Riande, J. Power Sources, 2007, 173, 68–76.
65. C. Guzman, A. Alvarez, O. E. Herrera, R. Nava, J. Ledesma-Garcia, L. A. Godinez, L. G. Arriaga and W. Merida, Int. J. Electrochem. Sci., 2011, 6, 4648–4666.
66. C. M. Wang, E. Chalkova, J. K. Lee, M. V. Fedkin, S. Komarneni and S. N. Lvov, J. Electrochem. Soc., 2011, 158, B690–B697.
67. E. Quartarone, A. Magistris, P. Mustarelli, S. Grandi, A. Carollo, G. Z. Zukowska, J. E. Garbarczyk, J. L. Nowinski, C. Gerbaldi and S. Bodoardo, Fuel Cells, 2009, 9, 349–355.
68. E. Quartarone, P. Mustarelli, A. Carollo, S. Grandi, A. Magistris and C. Gerbaldi, Fuel Cells, 2009, 9, 231–236.
69. E. B. Cho, H. Kim and D. Kim, J. Phys. Chem. B, 2009, 113, 9770–9778.
70. N. J. Hao, L. L. Li, Q. Zhang, X. L. Huang, X. W. Meng, Y. Q. Zhang, D. Chen, F. Q. Tang and L. F. Li, Microporous Mesoporous Mater., 2012, 162, 14–23.
71. F. Lu, S. H. Wu, Y. Hung and C. Y. Mou, Small, 2009, 5, 1408–1413.
72. A. H. Lu, W. C. Li, W. Schmidt and F. Schuth, J. Mater. Chem., 2006, 16, 3396–3401.
73. T. Benamor, L. Michelin, B. Lebeau and C. Marichal, Microporous Mesoporous Mater., 2012, 147, 334–342.
74. F. Raji and M. Pakizeh, Appl. Surf. Sci., 2013, 282, 415–424.
75. A. S. Araujo and M. Jaroniec, Thermochim. Acta, 2000, 363, 175–180.
76. X. G. Wang, K. S. K. Lin, J. C. C. Chan and S. F. Cheng, J. Phys. Chem. B, 2005, 109, 1763–1769.
77. X. L. Xu, L. Li, H. Wang, X. J. Li and X. P. Zhuang, RSC Adv., 2015, 5, 4934–4940.
78. Y. H. Cui, A. P. Baker, X. Xu, Y. Xiang, L. Wang, M. Lavorgna and J. W. Wu, J. Power Sources, 2015, 294, 369–376.
79. M. Lavorgna, L. Sansone, G. Scherillo, R. Gu and A. P. Baker, Fuel Cells, 2011, 11, 801–813.
80. S. Meenakshi, A. K. Sahu, S. D. Bhat, P. Sridhar, S. Pitchumani and A. K. Shukla, Electrochim. Acta, 2013, 89, 35–44.
81. D. Cozzi, C. de Bonis, A. D'Epifanio, B. Mecheri, A. C. Tavares and S. Licoccia, J. Power Sources, 2014, 248, 1127–1132.
82. C. Y. Yen, C. H. Lee, Y. F. Lin, H. L. Lin, Y. H. Hsiao, S. H. Liao, C. Y. Chuang and C. C. M. Ma, J. Power Sources, 2007, 173, 36–44.
83. S. L. Zhong, C. G. Sun, Y. Q. Luo, W. C. Liu and S. Dou, J. Power Sources, 2013, 238, 485–491.
84. C. W. Lin, Y. F. Huang and A. M. Kannan, J. Power Sources, 2007, 171, 340–347.
85. P. N. Colomban and A. Novak, Proton Conductors. Solids, Membranes and Gels – Materials and Devices, 1992.
86. M. M. Hasani-Sadrabadi, I. Shabani, M. Soleimani and H. Moaddel, J. Power Sources, 2011, 196, 4599–4603.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra24210c

---