J. L. Reyes-Rodrigueza,
O. Solorza-Feriaa,
A. García-Bernabéb,
E. Giménezc,
O. Sahuquilloc and
V. Compañ*b
aDepartamento de Química – Centro de Investigación y de Estudios Avanzados del I.P.N., Av. IPN 2508, Col. San Pedro Zacatenco, 07360 México D.F., Mexico
bEscuela Técnica Superior de Ingenieros Industriales – Departamento de Termodinámica Aplicada, Universidad Politécnica de Valencia, Camino de vera s/n, 46020 Valencia, Spain. E-mail: vicommo@ter.upv.es; Fax: +34 96 387 79 24; Tel: +34 96 387 93 28
cEscuela Técnica Superior de Ingenieros Industriales – Departamento de Ingeniería Mecánica y de Materiales, Universidad Politécnica de Valencia, Camino de vera s/n, 46020 Valencia, Spain
First published on 31st May 2016
Nanofiber mats of SPEEK70wt%–PVB30 wt% (polyvinyl butyral)-based composite membranes were prepared by varying the electrospinning time in order to obtain mats with different thicknesses. These mats were embedded in SPEEK65wt%–PVA35wt% (polyvinyl alcohol) polymer solution to fill the pores in the fibers. The obtained membranes with different mat thicknesses have been characterized by water uptake, ionic exchange capacity, scanning electron microscopy, mechanical properties and proton conductivity. Microtensile test studies reveal that the maximum tensile strength increases as the thickness of the SPEEK–PVB nanofiber mats increases, resulting in more flexible composite membranes compared to a pure SPEEK–PVA membrane obtained by casting. The proton conductivity occurs more easily through the nanofiber than through the matrix phase, and the best conductivity (0.038 S cm−1) was measured at 120 °C for the composite membrane of SPEEK–PVB nanofiber mats obtained after 12 hours of electrospinning time. This value suggests that our composite membranes have high potential to function in the temperature range between 100 and 140 °C without losing their strength and while maintaining their high proton conductivity, making them an excellent candidate for fuel cells that operate at intermediate temperatures.
However, efficient use of Nafion membranes is limited to a low temperature range below their glass transition temperature (Tg), in the vicinity of 100 °C, because their performance decreases considerably as a result of low proton conductivity due to dehydration at higher temperatures; this limits the maximum operating temperature of Nafion membranes to 80 °C.5,6 In contrast, the advantages of operating PEMFCs at high temperatures (100 to 200 °C) include several factors, such as: (1) increased tolerance to CO2 poisoning of Pt-based noble catalysts by using hydrogen from the reforming process and/or air as the oxygen source, (2) avoidance of the necessity of cooling systems and potential use of thermal energy for a co-generation process, (3) avoidance of the use of humidification systems in the gas supply to maintain hydration, (4) increased diffusion rate of the charge carriers of the membrane, (5) accelerated kinetic reaction of the catalytic materials with the reaction gases.5–12 For these reasons, high temperature fuel cells (HT-PEMFC) are considered to be a potential solution to address the current technical challenges of low temperature fuel cells (LT-PEMFC) and may be an ideal alternative, at least, for automotive purposes.5,8,9
Significant efforts have been made to search for new polymeric materials with proton exchange capacity, such as modified perfluorosulfonic acid polymer membranes, alternative sulfonated polymers, composite membranes, and acid–base polymer membranes.8,13–18 For more details and information, we recommend the work of Bose et al.,6 where current trends and challenges in new polymer membranes are addressed. Among these, one of the most studied systems for HT-PEMFC applications is H3PO4 doped polybenzimidazol (PBI), reported by Savinell and co-workers.19–23 Proton conduction through the solid PBI matrix is less dependent on water content than that of the perfluorosulfonic system,16,19,23 which allows these membranes to be operated at 100 to 200 °C with low relative humidity. However, the disadvantage of using acid doped PBI membranes is that the typical Pt based catalyst materials tend to suffer agglomeration and dissolution under high temperature and in the acidity of the H3PO4 environment.17,23
Our research group has been focusing on SPEEK polymer,24–26 which has been extensively studied by numerous research groups due to its high thermal (Tg > 170 °C) and chemical stabilities, availability of materials and low cost for application as a precursor of protonic exchange membranes for intermediate temperature (100 to 150 °C) PEMFCs and DMFCs.5,6,27–33 The main drawback of SPEEK has been found to be excessive swelling and dissolution in hot water, which prevents its practical use in DMFCs at intermediate temperatures24,26 but not in the case of PEMFCs fed with hydrogen. Crosslinking SPEEK with other polymers enhances its mechanical stability under high temperature conditions.26,28,30,32 Immersion of the membranes in boiling water for 1 h is the selected examination procedure.24
In our previous work, we observed that the nanofiber morphology exhibits a large surface area which can be functionalized and modified to tailor the interface properties; nanocomposite membranes were prepared based on (i) a Nafion® matrix with poly-vinyl alcohol (PVA) nanofibers (Nafion/PVA) and (ii) a blend based on poly-ether-ether-ketone sulfonated with PVA (SPEEK–PVA) with nanofibers of SPEEK blended with polyvinyl butyral (SPEEK–PVB) to obtain the composite membrane (SPEEK–PVA/SPEEK–PVB). The former is intended for DMFC applications below 80 °C and the latter for operation within the 80 to 140 °C range. From the characterization of the obtained nanocomposite membranes, it is revealed that the nanofibers of SPEEK/PVB have higher proton conductivity than those prepared by the casting method. The incorporation of the nanofiber mats into the SPEEK–PVA matrix provided mechanical stability and some proton conductivity up to a crosslinking temperature of 120 °C.24,25 As a result, mechanical reinforcement of the composite membranes leads to lower water uptake and swelling values, which causes the methanol permeability to decrease in DMFC conditions.24–26 In these studies, the surface of the PVA nanofibers was functionalized with sulfonic acid moieties through the reactivity of the OH groups of PVA via an acetal formation reaction. The interface compatibility was attributed to hydrogen bonding between the sulfonic acid species (–SO3H). The PVA nanofiber mats also provided a strong mechanical reinforcing effect to the composite membranes, which resulted in the successful preparation of low thickness membranes (<20 μm). Additionally, low thickness membranes with reduced ohmic losses can be fabricated.34 Generally, thin membranes are preferred for the operation of hydrogen fuel cells at lower relative humidity levels because they facilitate water transport between the cathode and the anode.
Electrospinning is a suitable and standard technique for the production of nanofibers made from a broad selection of materials, although new, more productive methods are being developed.34–43 In this sense, electrospun nanofibers are receiving considerable attention in the fabrication of novel proton exchange membranes with enhanced mechanical, chemical and proton conductivities36–41 where the conduction mechanism can occur more efficiently through the surface of the nanofibers.43
From the experience collected from our previous studies, in this work we offer a new strategy to obtain composite membranes for use in fuel cells operating in the range of 120 to 140 °C. In order to attain this goal, we follow our previous strategy of preparing composite membranes of hydrophilic PVA and hydrophobic PVB blended with SPEEK with different nanofiber thicknesses obtained by varying the electrospinning time. The corresponding membranes were crosslinked and characterized by water uptake, ionic exchange capacity, scanning electron microscopy (SEM), mechanical properties and proton conductivity, in order to probe which composition of nanofiber mat and polymer matrix was the best combination to build membrane-electrode assemblies.24–26
The sulfonation reaction is an electrophilic substitution that occurs at certain active sites of the PEEK structure where the electron density favors substitution. The electrophilic substitution can take place in one of the four available positions of the aromatic ring flanked by ether groups of the monomer unit (first type substitution). It is not possible to carry out a sulfonation process under normal conditions on the aromatic rings linked by the carboxylic group due to its electron withdrawing nature, which reduces the electron density on those sites. However, by increasing the temperature and extending the reaction times, it is possible to perform substitutions in the other two aromatic rings (second type substitution).5,29 The reaction was carried out at 55 °C for 12 hours of continuous sulfonation to prepare SPEEK with a high degree of sulfonation, taking into consideration our laboratory conditions.
According to previous reports of our research group,25,26 a blended solution of SPEEK and PVB was prepared in DMAc solvent maintaining a ratio of 70/30 w/w, respectively, as follows: 14 g of dry SPEEK and 6 g of PVB were dissolved in DMAc at 80 °C under magnetic stirring until complete homogenization (∼1 h). A calculated volume of DMAc was added to obtain a concentration of 20 wt% of the total polymer mass (SPEEK–PVB) with respect to the solvent mass. A concentration lower than 17.5 wt% leads to a low solution viscosity and causes defects, known as beads, in the fiber structure. Therefore, the optimum working range is between 17.5 and 20 wt%. Due to the environmental conditions of humidity and temperature in our laboratory, the SPEEK–PVB solution was always maintained under magnetic stirring before being electrospun. Even the syringe was replaced with fresh polymer solution after a certain amount of time to prevent phase separation of the components. Fiber mats of 20 wt% SPEEK70-PVB30/DMAc solution were electrospun using a Super ES-2 model E-Spin Nanotech electrospinning machine. The polymer solution was injected by a syringe pump with a flow rate of 0.2 ml h−1 to the tip of a steel capillary which was separated by 18 cm from the rotating drum collector (1200 rpm). A potential difference of 20 kV was applied between both electrodes. The electrospinning time was varied from 2 to 14 hours (every 2 h) with the purpose of obtaining different mat thicknesses. The obtained nanofiber mats were dried in an oven first at 80 °C for 3 h and then at 160 °C for 30 minutes in order to remove residual DMAc solvent. The mats showed hydrophobic behavior in the presence of water droplets but were practically soluble when immersed in a mixture of ethanol and water. Cross-linking at 180 °C for 2 h resolved this problem, providing high chemical stability to the fibers, a property that is remarkable for applications in methanol or ethanol fuel cells. In this step, a color change from white to brown was observed.
Membrane | M-02 | M-04 | M-06 | M-08 | M-10 | M-12 | M-14 |
Electrospinning time (h) | 2 | 4 | 6 | 8 | 10 | 12 | 14 |
SPEEK70–PVB30 nanofiber mat thickness (μm) | 9 | 15 | 29 | 47 | 62 | 114 | 140 |
SPEEK–PVB–PVA composite membrane thickness (μm) | 112 | 116 | 124 | 134 | 150 | 156 | 168 |
For comparison, a pure SPEEK–PVA membrane was prepared by casting following the same methodology described above using the 12 wt% SPEEK65–PVA35/H2O polymer solution and the Teflon frames as the cast.
![]() | (1) |
The swelling degrees, both in-surface and in-thickness, were calculated by similar expressions considering the wet surface area (A) and/or wet thickness (L) (under the same hydration conditions mentioned above) of the composite membrane samples with respect to the dry conditions, according to eqn (2) and (3):
![]() | (2) |
![]() | (3) |
The ion exchange capacity (IEC) was estimated from samples (∼100 to 200 mg) of the pristine SPEEK polymer, 12 h SPEEK70–PVB30 nanofiber mat and SPEEK–PVB–PVA composite membranes which were immersed in 50 ml of 1 M NaCl solution and stored in sealed bags for 3 days. During this time, the protons liberated from the SPEEK polymer chains interacted with the Cl− anions of the salt to form HCl by an ionic exchange mechanism. The solution was filtered and then titrated in triplicate with a 0.0095 M NaOH standard solution. The IEC (meq g−1) was calculated according to eqn (4):
![]() | (4) |
![]() | (5) |
The SPEEK–PVB nanofiber mat sample with 12 h of electrospinning time was chosen for physical characterization because, as will be seen later, it had higher proton conductivity when embedded in the final membrane and therefore is of greater interest.
From the frequency dependence of the complex impedance Z*(ω) = Z′(ω) + jZ′′(ω), the real part of the conductivity is given as
![]() | (6) |
The surface and the cross-section of the final SPEEK–PVB–PVA composite membrane are shown in Fig. 1c. From the micrographs, it is possible to observe that the nanofibers are properly embedded in the SPEEK65–PVA35 aqueous polymer matrix, mostly in the center of the membrane (1c–d). Furthermore, a compact membrane surface with minimal defects was observed, and no voids or pores with exposed fiber were observed (Fig. 1e).
The above observations confirmed that with our preparation method, it is possible to obtain membranes with sandwich structures with adequate penetration and impregnation of SPEEK–PVA polymer solution into the SPEEK–PVB nanofiber mat.
Table 2 shows the values obtained for water uptake, swelling, ion exchange capacity and sulfonation degree for all composite membranes, pristine SPEEK and SPEEK–PVB nanofiber mats. It is observed that the pristine SPEEK polymer obtained after 12 h of sulfonation has an IEC of 2.35 meq g−1, which represents a sulfonation degree of 83%, a high value when compared with commercial products ranging from 1.75 to 2.05 meq g−1. However, the IEC value of the SPEEK–PVB nanofibers decreases drastically to 0.60 meq g−1 (SD = 18%) when electrospun; this may be due to the presence of PVB in the crosslinking treatment applied to the fiber mats. When SPEEK–PVB is embedded with the SPEEK–PVA solution (SPEEK–PVB–PVA composite membranes), the IEC value increases to 1.58 to 1.71 meq g−1 (SD = 52 to 57%). We believe that SPEEK–PVA is responsible for the increase in IEC as a consequence of its higher sulfonation degree, influenced by the PVA OH– groups in the SPEEK chain. Previous studies carried out by our group have shown that the IEC of SPEEK–PVA membranes is around 0.22 meq g−1.24–26 A slight downward trend is clearly observed in Table 2 in the value of IEC as the membrane thickness increases. The final membrane thickness is influenced by the thickness of the fiber mat obtained by varying the electrospinning time. In M-02, the thickness of the fibers was 9 μm, resulting in lower proportions of SPEEK–PVB with respect to the proportion of SPEEK–PVA polymer in the membrane; therefore, the IEC value will be slightly larger compared to that of M-14, with a fiber thickness of 140 μm, which has a higher proportion of SPEEK–PVB in the membrane and would result in a decreased IEC value for this membrane. For water uptake and swelling, the data do not show a meaningful trend.
Sample | Membrane thickness (μm) | Water uptake (%) | In-plane swelling (%) | Thickness swelling (%) | IEC (meq g−1) | SD (%) |
---|---|---|---|---|---|---|
Pristine SPEEK | — | — | — | — | 2.35 ± 0.00 | 83 ± 1 |
12 h SPEEK–PVB nanofiber mat | — | — | — | — | 0.60 ± 0.00 | 18 ± 0 |
M-02 | 112 | 238 ± 7 | 87 ± 8 | 34 ± 2 | 1.71 ± 0.00 | 57 ± 0 |
M-04 | 116 | 116 ± 6 | 49 ± 8 | 15 ± 1 | 1.72 ± 0.00 | 57 ± 0 |
M-06 | 124 | 158 ± 3 | 69 ± 0 | 78 ± 3 | 1.68 ± 0.00 | 56 ± 0 |
M-08 | 134 | 148 ± 3 | 27 ± 6 | 66 ± 3 | 1.68 ± 0.02 | 56 ± 1 |
M-10 | 150 | 135 ± 9 | 21 ± 0 | 89 ± 5 | 1.67 ± 0.04 | 56 ± 1 |
M-12 | 156 | 141 ± 7 | 47 ± 0 | 48 ± 6 | 1.60 ± 0.00 | 53 ± 0 |
M-14 | 168 | 161 ± 11 | 42 ± 8 | 45 ± 2 | 1.58 ± 0.00 | 52 ± 0 |
Sample | Membrane thickness (μm) | Young's modulus E (GPa) | Maximum tensile strength σmax (MPa) | Maximum strain εmax (mm) | Tenacity (MJ m−3) |
---|---|---|---|---|---|
12 h SPEEK–PVB fiber mat | 170 | 0.1 ± 0.0 | 6.2 ± 0.1 | 0.2 ± 0.0 | 1.0 ± 0.1 |
M-02 | 112 | 0.9 ± 0.1 | 22.1 ± 0.8 | 0.2 ± 0.0 | 2.1 ± 0.3 |
M-04 | 116 | 1.2 ± 0.1 | 24.8 ± 1.4 | 0.2 ± 0.0 | 1.8 ± 0.3 |
M-06 | 124 | 1.4 ± 0.1 | 32.1 ± 5.8 | 0.2 ± 0.0 | 3.5 ± 1.6 |
M-08 | 134 | 1.1 ± 0.2 | 22.3 ± 3.1 | 0.2 ± 0.0 | 1.5 ± 0.1 |
M-10 | 150 | 1.7 ± 0.2 | 44.6 ± 2.1 | 0.3 ± 0.0 | 6.4 ± 2.4 |
M-12 | 156 | 1.7 ± 0.1 | 49.4 ± 6.2 | 0.3 ± 0.1 | 4.5 ± 0.5 |
M-14 | 168 | 1.7 ± 0.1 | 33.9 ± 6.0 | 0.1 ± 0.0 | 1.5 ± 0.8 |
SPEEK–PVA membrane | 110 | 2.7 ± 0.5 | 66.9 ± 3.9 | 0.2 ± 0.1 | 10.2 ± 2.4 |
From the data in Table 3, it can be clearly seen that there are two limits: the maximum tensile strength in the SPEEK–PVB nanofiber sample (6.2 MPa) and the maximum tensile strength in the SPEEK–PVA sample (66.9 MPa). The results for all the samples of SPEEK–PVB–PVA composite membranes are within these limit values, and we can see an increase in the tensile strength as the membrane thickness is increased (with consequent thickening of the fiber mat through increasing electrospinning time). This trend is best seen in Fig. 2b. This means that the amount of SPEEK–PVB fibers in the composite membranes seems to have very little influence on their tensile strength. However, the incorporation of the fiber promotes increased stretch (high elasticity) before rupture, especially for fibers of longer electrospinning time (10 to 12 h), compared with SPEEK–PVA pure membrane, which breaks abruptly (Fig. 2a). We can observe that the M-08 and M-14 membranes do not follow the same upward trend as the other membranes. We theorize that this is due to structural defects obtained during the preparation of the membranes and to water effects (the water uptake percentage and the degree of swelling that these membranes acquired).
Fig. 3 shows the double logarithmic plots of the imaginary permittivity ε” versus the frequency for the M-04 sample at all temperatures. The M-04 membrane was chosen from among all membranes because it can be a typical membrane for fuel cell applications, considering that it has a total thickness of 116 μm composed of a nanofiber phase with a thickness of 15 μm and a bulk phase approximately 101 μm in thickness. From Fig. 3, we can see that the behavior of the SPEEK–PVB/SPEEK–PVA samples is linear in the region of high frequency, with a slope independent of the amount of nanofiber loading. The slope of the straight line is practically −1 for all the temperatures studied, with a correlation coefficient of 0.999. This is the typical contribution to the dielectric loss from electrical conduction, which means that the behavior of the membranes for this interval of frequencies is that of a pure conductor.49 From the intercepts of the straight line observed at the high frequency region, shown in Fig. 3, we can obtain the values of the DC conductivity for the nanocomposites for each temperature. For the rest of the membranes, similar behavior is observed, as shown in the ESI (Fig. S9–S15†). The values obtained for all the membranes are given in Table 4.
T (°C) | σNyquist × 103 (S cm−1) | σBode × 103 (S cm−1) | σDC × 103 (S cm−1) | D+ × 106 (cm2 s−1) | Eact (kJ mol−1) | σDC × 103/cion (S kg cm−1 eq−1) |
---|---|---|---|---|---|---|
Sample: M-02; L: 112 μm | ||||||
80 | 15.5 | 15.1 ± 0.4 | 11.4 ± 2.1 | 2.8 | 22.4 ± 2.9 | 6.67 |
100 | 17.8 | 17.7 ± 0.4 | 14.5 ± 2.8 | 3.5 | 8.48 | |
120 | 17.6 | 17.5 ± 0.3 | 16.1 ± 3.1 | 3.6 | 9.42 | |
140 | 16.2 | 16.4 ± 0.1 | 19.9 ± 2.4 | 3.5 | 11.64 | |
![]() |
||||||
Sample: M-04; L: 116 μm | ||||||
80 | 19.4 | 19.6 ± 0.2 | 19.7 ± 3.9 | 3.6 | 21.0 ± 3.4 | 11.45 |
100 | 21.7 | 21.5 ± 0.2 | 22.9 ± 4.4 | 4.2 | 13.31 | |
120 | 18.5 | 18.7 ± 0.1 | 24.3 ± 3.1 | 3.8 | 14.13 | |
140 | 12.7 | 12.9 ± 0.1 | 19.6 ± 1.0 | 2.8 | 11.40 | |
![]() |
||||||
Sample: M-06; L: 124 μm | ||||||
80 | 9.4 | 9.3 ± 0.2 | 7.7 ± 0.6 | 1.8 | 28.6 ± 3.8 | 4.58 |
100 | 14.6 | 14.4 ± 0.2 | 13.4 ± 1.0 | 2.9 | 7.98 | |
120 | 17.9 | 17.9 ± 0.3 | 13.1 ± 2.0 | 3.7 | 7.80 | |
140 | 15.2 | 15.3 ± 0.2 | 13.4 ± 1.5 | 3.4 | 7.98 | |
![]() |
||||||
Sample: M-08; L: 134 μm | ||||||
80 | 11.2 | 11.3 ± 0.0 | 18.2 ± 0.5 | 2.1 | 20.9 ± 3.5 | 10.83 |
100 | 14.2 | 14.1 ± 0.1 | 20.4 ± 1.1 | 2.8 | 12.14 | |
120 | 14.5 | 14.4 ± 0.1 | 20.8 ± 1.2 | 3.0 | 12.38 | |
140 | 12.9 | 12.9 ± 0.1 | 19.0 ± 0.9 | 2.8 | 11.31 | |
![]() |
||||||
Sample: M-10; L: 150 μm | ||||||
80 | 16.5 | 16.2 ± 0.1 | 22.0 ± 1.5 | 3.1 | 23.3 ± 3.1 | 13.17 |
100 | 19.5 | 19.6 ± 0.1 | 27.1 ± 2.1 | 3.9 | 16.23 | |
120 | 20.8 | 20.0 ± 0.1 | 29.1 ± 1.9 | 4.2 | 17.43 | |
140 | 18.5 | 18.7 ± 0.1 | 27.7 ± 1.4 | 4.1 | 16.59 | |
![]() |
||||||
Sample: M-12; L: 156 μm | ||||||
80 | 29.2 | 30.0 ± 0.5 | 30.6 ± 4.4 | 5.9 | 28.7 ± 4.3 | 19.13 |
100 | 38.2 | 37.1 ± 0.7 | 32.5 ± 7.1 | 7.7 | 20.31 | |
120 | 38.2 | 38.2 ± 0.5 | 34.9 ± 7.7 | 8.4 | 21.81 | |
140 | 35.5 | 35.1 ± 0.4 | 35.7 ± 6.1 | 8.1 | 22.31 | |
![]() |
||||||
Sample: M-14; L: 168 μm | ||||||
80 | 24.3 | 24.7 ± 0.2 | 33.3 ± 2.9 | 4.9 | 28.4 ± 3.3 | 21.08 |
100 | 30.4 | 30.9 ± 0.3 | 36.7 ± 5.5 | 6.5 | 23.23 | |
120 | 31.5 | 31.9 ± 0.2 | 41.6 ± 5.8 | 7.1 | 26.33 | |
140 | 28.6 | 28.7 ± 0.2 | 38.4 ± 4.0 | 6.7 | 24.30 | |
![]() |
||||||
Sample: pure SPEEK–PVA membrane; L: 134 μm | ||||||
80 | 2.8 | 2.4 ± 0.1 | 2.7 ± 0.4 | 1.3 | 33.4 ± 2.2 | 4.66 |
100 | 0.6 | 0.5 ± 0.0 | 0.4 ± 0.1 | 0.3 | 0.69 | |
120 | — | — | — | — | — | |
140 | — | — | — | — | — |
To probe whether the conductivity obtained is the true conductivity of the composites, we have represented the Bode50 plot, shown in Fig. 4, where the real part of the conductivity, σ′, and the phase angle (°) are represented as a function of the frequency for the M-12 sample. We can see on this figure that the real part of the conductivity, σ′, reaches a constant value (plateau), and the phase angle (°) reaches a maximum close to zero in the high frequency region. Similar plots were found for the other samples (Fig. S1–S8 in the ESI†).
In order to obtain exact values of σdc, the impedance data in terms of conductivity were fitted by a Nyquist plot using the equivalent circuit shown in the inset of Fig. 3. In the equivalent circuit, the bulk resistance of the membrane, represented by R0, is associated in series with a circuit composed of a resistance Rp (representing a polarization electrode resistance) in parallel with a constant phase element (CPE) of admittance Y* = Y0(jωτ)n, where j is the imaginary unit (j2 = −1), n is the exponent independent of frequency lying in the range 0 < n < 1, ω (=2πf) is the angular frequency and C is a CPE-parameter which represents the differential capacitance of the interface when n = 1, a resistor when n = 0 and a inductor in the case of n = −1; however, when n < 1, C cannot represent the capacitance because its units are not simply Farads.51,52 This CPE accounts for the interfacial phenomena in the membrane-electrode interface. In this region, the Nyquist53 diagrams, −Z′′ (imag.) vs. Z′ (real), are semicircles intersecting the abscissa axis in the high frequency region at Z′ = R0; this value permits us to obtain the DC conductivity. Semicircle-like waves are observed experimentally as a result of polarization processes and other possible phenomena taking place at the membrane electrode interface. In the cases where the systems combine polarization processes with a resistance independent of the frequency or ohmic resistance, R0, the equivalent electrical circuit can be given by a series of Rpi–CPE parallel circuits in series with R0; such a case is represented in the inset of Fig. 3, where the complex impedance of the system is given by
![]() | (7) |
A close inspection of Fig. 4 shows that the values of the real part of the conductivity reach a plateau when the phase angle tends to zero. In our samples, we can observe two plateaus: one situated in the region of high frequencies associated with the protonic conductivity in the bulk SPEEK–PVA pure membrane, identical to the bulk conductivity σdc, reflecting long-range proton transport, and the other in the region of low frequencies that can be associated to the proton conductivity trough of the SPEEK–PVB nanofibers and also to the electrode polarization via the formation of electrochemical double layers and polymer relaxation.54–57 Between the regions of high and low frequency, there is a decrease in the conductivity that can be explained as a Debye relaxation due to the macroscopic polarization of the ionic charges as a consequence of the applied electric field. This relaxation is characterized by a relaxation time which is dependent on the temperature, the chemical structure of the membrane and its thickness.58–60 The value of the conductivity obtained in the region of high frequencies where the phase angle reaches zero, as previously stated, represents the proton conductivity of the membrane. In Table 4, we can see the values obtained for each of the composite membranes and for the pure SPEEK–PVA membrane for comparison.
In the inset of Fig. 4, we plotted the Nyquist diagram for the same temperatures represented in the Bode diagram of conductivities. In this figure, the real and imaginary components of the complex impedance are plotted for the entire range of frequencies. We can observe for all the composite membranes and for all temperatures two semicircles, intersecting the abscissa axis in the high frequency region at Z′ = R0 (i.e. the membrane resistance). Semicircle-like waves are observed as a result of polarization processes, the two possible relaxation processes associated with the bulk and nanofiber components, and other phenomena taking place in the system composed of the SPEEK/PVB nanofiber–SPEEK/PVA bulk-electrode interface.
To appreciate the existence of the two relaxation processes associated with the SPEEK–PVB nanofiber and the SPEEK–PVA bulk observed from the Bode plots, we have also analyzed the experimental results from the Nyquist plot shown in the inset of Fig. 3, corresponding to the fit –Z′′ vs. Z′. The solid line that appears in the inset corresponds to the best fit of −Z′′ vs. Z′ to the experimental values previously represented by symbols at each temperature in the Nyquist plot inside the Bode diagram. In Table 4, we gather all the membrane conductivities obtained by Bode and Nyquist plots at several temperatures.
The conductivity values change very little when we compare the method used to estimate them at each temperature, and they increase by about 100 to 1000 times when the amount of nanofiber increases. From these results, we can observe that the proton conduction mechanism is more efficient through the surface of the SPEEK–PVB nanofibers than through the matrix of SPEEK–PVA; similar results have been observed in other nanocomposites.61,62 On the other hand, it is possible to observe a critical thickness as a function of the electrospinning time (M-12 with 12 hours of electrospinning time of SPEEK–PVB) where the protonic conductivity is the highest, as shown in Fig. 5a.
A similar comparison of the protonic conductivity values as a function of the temperature analyzed by EIS is shown in Fig. 6, where the SPEEK–PVB–PVA M-12 composite membrane showed the highest values at 120 to 140 °C. The conductivity of the composite membrane M-14 at 140 °C is about 0.03 S cm−1, quite similar to Nafion membranes under DMFC operation at 70 °C.63 This result is noteworthy because it shows that these new composite membranes can be applied in this temperature range while maintaining good conductivity compared with Nafion membranes, which are susceptible to degradation.8,9
![]() | ||
Fig. 6 Conductivity comparison for all the SPEEK–PVB–PVA composite membranes obtained as a function of the temperature, analyzed by EIS. |
Proton transport in acidic membranes is a complex process involving diverse steps: the dissociation of the proton from the fixed sulfonic acidic groups and its transfer through the SPEEK–PVB nanofibers and the SPEEK–PVA matrix. Taking into account that the measured conductivity is due to the contribution of all the charge carriers and knowing that these are disseminated both through the nanofiber phase (SPEEK–PVB) and through the matrix phase (SPEEK–PVA bulk), we can observe, in a first approximation, that the total resistance to the proton transport can be given as:
![]() | (8) |
Taking into account eqn (6), we have:
![]() | (9) |
Assuming that the area is the same for the phases, bulk and nanofiber, we have in first approximation that:
![]() | (10) |
We determined Lnanofiber and Lmembrane experimentally; consequently, we can give an estimate of the thickness of the matrix phase. Furthermore, we have determined the conductivity of the composite membrane and of the membrane without nanofiber. Thus, we can approximately estimate the contribution of the nanofibers to the conductivity of the composite membrane.
For example, it can be seen that for the M-08 membrane, at a temperature of 100 °C, with a nanofiber thickness of 134 μm and a thickness of the SPEEK–PVA bulk phase identical to the corresponding pure SPEEK–PVA membrane, it is readily apparent that the nanofiber has 10 times higher conductivity than the matrix phase. To do this, the conductivity value of the pure PVA–SPEEK membrane (0.5 × 10−3 S cm−1) given in Table 5 is considered. Similarly, for the M-12 membrane, the conductivity value calculated for the nanofiber (114 μm) is 2.67 × 10−2 S cm−1. This value is very similar to the value of 2.61 × 10−2 S cm−1 obtained in a previous study24 and indicates that our nanofibers have a conductivity about 50 times higher than the SPEEK–PVA phase. The results for all the composite membranes are shown in Table 5 and Fig. 6b, where comparisons of the conductivity for all the composite membranes as a function of the nanofiber mat thickness are given.
Composite membrane | σmembrane × 102 (S cm−1) | σnanofiber × 102 (S cm−1) | σnanofiber/σbulk |
---|---|---|---|
M-02 | 1.77 | 0.14 | 2.8 |
M-04 | 2.15 | 0.27 | 5.4 |
M-06 | 1.44 | 0.33 | 6.6 |
M-08 | 1.41 | 0.48 | 9.6 |
M-10 | 1.96 | 0.79 | 15.8 |
M-12 | 3.71 | 2.67 | 53.4 |
M-14 | 3.09 | 2.52 | 50.4 |
SPEEK–PVA pure membrane (bulk) | 0.05 |
Therefore, as an evident conclusion and fully justifying some approximations, we can say that conductivity occurs more easily through the nanofiber than through the SPEEK–PVA phase. These results are in agreement with the literature,62,64 where it is found that the proton conductivity of SPEEK nanofiber mat is nearly five times as high as that of the SPEEK composite membrane. Our result is also consistent with the work of Hongwei et al.,62 where two possible pathways are presented: (1) the long-range proton conductivity pathways are formed inside of the nanofibers,65 and (2) the long-range proton conductivity pathways are formed on the surfaces of the nanofibers.64 Conductivity may be improved through the bulk due to an induced orientation of the ionic channels along the nanofiber axis; otherwise, conductivity may be encouraged on the nanofiber surface through the strong interface formed between water molecules and external sulfonic groups.
The measure of the conductivity carried out in our work is transversal. The in-plane method only gives the conductivity value in the longitudinal direction (OX) or (OY); however, for applications in fuel cells, we need the transversal value (i.e. through the sample). To acquire this, we must sandwich a sample between two electrodes (gold electrodes in our study). The nanofibers were collected using a rotating drum collector where the flow rate conditions of SPEEK–PVB solution are injected in the direction of the cathode collector where practically all the nanofibers are aligned. Since the conductivity measurement is transversal, we can suppose that the protons move transversally to the sample (nanofibers). For this reason, we are in agreement with Hongwei's explanation: number 2 and number 1 would be unfavorable to proton transfer between nanofibers.
In Fig. 7, we plot the Arrhenius plot of the conductivity of the nanocomposite membranes in the temperature range from 20 °C to 140 °C except for the pure SPEEK–PVA membrane, where the temperature range was from 20 °C to 100 °C. From these results, we clearly observe two regions: a region of decrease in conductivity above 100 °C, with a bigger decrease in the case of pure SPEEK–PVA membrane, and another region at low temperature where Arrhenius behavior is observed for the conductivity. The difference between these two regions can be explained by taking into account that in the nanofiber phase (SPEEK–PVB), the charge transport takes place via a hopping mechanism between sulfonic groups in a frozen disorder structure.66,67 However, in the phase matrix composed of the polymer SPEEK–PVA, the diffusion of protons takes place both via a hopping mechanism and via H3O+ decrease by the effect of an increase in temperature due to the evaporation of water molecules from the polymeric matrix. Therefore, it is expected that higher values of activation energy will be associated to a higher amount of matrix phase, and smaller values should be expected with a higher amount of nanofiber phase. The activation energy is calculated in the interval where the Arrhenius plot presents a linear behavior. The obtained results are also given in Table 4, being in all cases smaller than the value obtained for the pure SPEEK–PVA membrane.
In binary systems, all the available ions participate in the ionic transport, since most of them may be interacting to form pairs or clusters; thus, it is very difficult to obtain the true contribution of each ion to the ionic transport. The conductivity is obtained for the complete system, and it represents the sum of the total contributions of the mobile charges, i.e., σ = σ+ + σ−. In studies to measure the diffusivity of ILs in polymerized ionic liquids (PILs) using the method of pulsed field gradient (PFG) NMR measurements, it has been observed that the conductivity of the PIL is greater than that of its molecular counterpart.66 This indicates that the cation mobility is very slow to provide a significant contribution to the true conductivity of the samples, and this should be dominated by the contributions of the anions. Thus, although all the available charges participate in the ionic transport, it can be assumed that the highest plausible contribution to the true conductivity, σ = σdc, for all the samples, is the mobility of the anions. Therefore, the diffusion coefficient can be calculated from the conductivity measurements, assuming that it is only associated with the contributions of the anions. From the experimental results of the conductivity associated with the protons (cations) and the ionic exchange capacity of the nanocomposite membranes, the diffusion coefficient has been estimated.54,68
![]() | (11) |
In Table 4, we show the values obtained for the proton diffusion coefficient in each of the composite membranes in the temperature range of 80 to 140 °C. Table 4 shows that the diffusivity of protons for each of the membranes increases as a function of the ionic concentration (protons) available in the membrane, varying with the relation and with the temperature. The same increment is observed as the membrane thickness increases because there is more protonic exchange polymeric material inside the membrane, as seen in Tables 1 and 2
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra08228a |
This journal is © The Royal Society of Chemistry 2016 |