Fuel cell tests only concern ADP-Morgane^® membrane. The increase of the SAFC working temperature is one of the major key for the development of this technology. So power density curves were recorded at different temperatures using ADP-Morgane^® non-treated membrane (Figure 10). The temperature increase from 24 °C to 70 °C leads to the enhancement of the maximal power density (from 3 mW cm⁻² at 24 °C up to 7.8 mW cm⁻² at 70 °C); the effect seems to be related to the improvement of reaction kinetics at the electrodes.

Figure 11 demonstrates that the use of argon plasma treated membranes (2 or 80 W, 10 min) results in the improvement of the cell electrical performance. Indeed, the comparison of results as a function of the plasma input power shows that, whether the membrane was modified on one or both side(s), the best performance is obtained with the lowest value of the plasma input power (i.e., 2 W). For the best plasma input power of 2 W, the plasma treatment on both sides of the membrane is more efficient (22.6 mW cm⁻²) than the treatment on only one side (18.5 mW cm⁻²). No significant difference has been detected between one and both side(s) treatment for the plasma input power of 80 W. When compared to the pristine ADP-Morgane^® membrane (7.8 mW cm⁻² at 70 °C), the maximum power density delivered by the best argon plasma treated membrane (2 W, 10 min, treatment on both sides of membrane) is almost three times higher (22.6 mW cm⁻² at 80 °C). The SAFC test of the best non-quaternized plasma deposited ADP-Morgane^® membrane (TAA deposit at P_w = 40 W, DC = 10% and τ = 20 min) also revealed better performance than the untreated membrane (maximum power density equal to 21 mW cm⁻² at 80 °C). These noticeable improvements are consistent with the conductivity, water content, IEC and glycerol retention enhancements previously mentioned.

In order to better understand the effect of the plasma treatment on the enhancement of the fuel cell performance, polarization curves have also been registered. Figure 12 illustrates the case of the one-side argon plasma treatment (2 W, 10 min) which is representative of all other plasma treatments. It is shown that there is no effect of the plasma treatment in the activation region (low current density and high cell voltage region), whereas there is one in the ohmic region (linear part of the polarization curves), which is certainly related not only to the difference in membrane conductivity between untreated and treated membranes but also to the difference in membrane-electrode interface conductivity. Indeed, the effect of membrane surface etching inherent to plasma treatment may have increased both the adherence of the membrane on the electrode and the surface area contact between the membrane and the electrode, favoring the hydroxyl ions transfer and mimicking mass transfer enhancement at high current densities.

Plasma process was performed in a 30 L commercial capacitively coupled plasma reactor built by MECA2000 (depicted in Figure 13), pumped through a turbomolecular pump (Adixen Pascal 2015 SD coupled with a Boc Edwards EXC 120 module) for pre-treatment high vacuum and a primary pump (Adixen Pascal 2015 SD) for the plasma treatment or deposition.

A liquid nitrogen trap was positioned between the plasma chamber and the primary pump in order to trap the precursor and fragments vapors in the case of plasma deposition. The pressure inside the chamber was sensed and monitored respectively by a Leybold Vakuum Ceravak gauge and Edwards Speedivalve valve. A 13.56 MHz radiofrequency (RF) source (Dressler CESAR 136) was used to supply power to the parallel and vertical plate electrodes (gap between both electrodes: 2 cm). The RF and grounded electrodes were disk electrodes (10 cm diameter), being respectively immovable and rotative (speed ~6 rpm). Argon (Air liquid, purity > 99.999%), used as main gaseous phase for plasma treatment or as carrier gas for plasma deposition, was fed into the system via a UNIT Instruments Inc. (UFC1000) gas flow meter. For plasma deposition, the precursor triallylamine (Aldrich, 99%), named TAA in this paper, was contained in a flask maintained at a temperature of 45 °C. The stainless steel pipe between the precursor flask and the reactor was maintained at a temperature of 60 °C, to avoid precursor condensation in the pipe.

This section describes a standard procedure for plasma treatment using argon as gaseous phase (on both ADP-Morgane^® and AMELI-32^®) and plasma deposition using triallylamine as precursor (on ADP-Morgane^® only). First, a piece of membrane (from 20 × 25 up to 30 × 30 mm²) and a silicon wafer (from 10 × 20 up to 30 × 30 mm²) were fixed, using Kapton tape, on the removable holder, which was then positioned on the anode port. The plasma chamber was then evacuated using the primary vacuum system until approaching the limit pressure (P < 10⁻² mbar) and further evacuated for at least one hour at this pressure. The rotation of the anode was started, before switching to the secondary vacuum system. In the case of plasma deposition, the liquid nitrogen trap was filled and its level regularly adjusted for all the process long. The chamber was then evacuated until reaching a pressure of ca. 10⁻⁵ mbar and the vacuum was maintained at this pressure for at least 15 min. The system was then switched back to the primary vacuum setup and the argon flow started and set to a mass flow opening of 30% or 1.7%, for plasma treatment or deposition respectively. Once the internal pressure of argon in the reactor reached equilibrium, the plasma glow discharge was switched on and maintained for the desired duration in the case of plasma treatment. In the case of plasma deposition, the injection pipe system was flipped to the argon-TAA mixture; once an equilibrium pressure of ca. 0.9–1.3 × 10⁻¹ mbar was reached, the plasma glow discharge was switched on and maintained for the desired duration. After the plasma discharge was switched off, the argon stream was set to a mass flow of 50% and maintained for half an hour (after flipping back of the injection pipe system to the sole argon line in the case of plasma deposition). After stopping the argon flow, the evacuation was further continued using the primary vacuum system for at least 30 min, before returning the chamber to atmospheric pressure and recovering the samples.

The parametric study consisted in varying the discharge power (P_w) from 2 to 100 W (for argon plasma treatment) or from 20 to 250 W (for plasma deposition); the duty cycle (DC, defined as the ratio of the effective on light plasma duration to the overall plasma process duration) was equal to 10% or 100% and the plasma glow duration (τ) was fixed at 10 or 20 min (for argon plasma treatment) or varied from 2 to 60 min (for plasma deposition). The average input power, defined as P_w × DC, is noted P_A.

In the case of argon plasma treatment, membranes were treated on one or both side(s); in the latter case, the plasma treatment procedure was implemented twice, reversing the membrane on the substrate holder between both procedures. In the case of plasma deposition, some films were post-treated by methyliodide (at 85 °C for 16 h in a 10:90 v/v methyliodide/acetonitrile mixture) for the quaternization of the amine functions.

SEM measurements were conducted on a Hitachi S4500 instrument under an excitation voltage value between 2 and 8 kV depending on each sample’s surface charging. Before analysis, pieces of membranes were immersed in liquid nitrogen and broken in order to have a neat cut of the membranes’ cross-section; then pieces of membrane were stuck on double face tape on microscope support and Pt-metalized by sputtering under vacuum.

XPS analyses were conducted to depict the chemical composition of plasma films (deposited on silicon wafer). They were performed on an SIA 200 instrument from Riber Cameca UHV or on an ESCALAB 250 instrument from Thermo Electron. The source of excitation was the Al K_α X-ray at 1486.6 eV. The photoelectron spectra were calibrated in bond energy to the C–C/C–H bonds component of the carbon C_1s peak at 284.8 eV. Samples were analyzed both at their surface and in the bulk. The bulk was characterized after erosion of the deposits (using an argon ionic beam) for etching durations specifically adjusted according to, on the one hand, film thickness and hardness, and on the other hand, apparatus and energy of the argon ionic beam. Optimized etching durations were 600 s in the case of non-quaternized films bombarded with a 0.6 kV argon ionic beam on the SIA 200 instrument; they were 135 s and 155 s for non-quaternized and quaternized films, respectively, bombarded with a 2 kV argon ionic beam on the ESCALAB 250 instrument.

IEC measurements were performed using a Tacussel TT-2 auto-titrator coupled to a Radiometer ABU901 auto-burette. Titration was based on the dosage of Cl⁻ anions delivered from a Cl⁻-membrane exchanged in a nitric acid solution. Typically, a piece of 1 cm² of membrane was immersed in ca. 25 mL of HCl 1 mol L⁻¹ under stirring during 1 h. The process was repeated once before recovering the piece of membrane, which was then copiously rinsed with ultrapure water (R ≤ 18.2 MΩ cm) before being immersed in a 50 mL solution of 1 mol L⁻¹ HNO₃ (volume precisely determined) for at least 12 h. At the end of this anion-exchange period, the piece of membrane was recovered, copiously rinsed with ultrapure water and retained for further analysis of the water content. The nitric solution was then titrated using a 10⁻³ mol L⁻¹ silver nitrate solution and the Cl⁻ anion concentration of the anionic-exchange nitric acid solution determined (C_Cl). The IEC value was calculated using the following formula: IEC = (C_Cl × V_HNO3)/m_d, where V_HNO3 is the volume of the anionic-exchange solution (50 mL) and m_d is the weight of the dried piece of membrane (determined during the water content analysis). Titration was performed at least three times on each anionic-exchange nitric acid solution.

Water content of the membranes was determined using a Mettler Toledo HR53 moisture analyzer. Typically, the 1 cm² piece of membrane retained during the IEC determination was immersed in 25 mL ultrapure water for 1 h, then recovered and immersed in ca. 25 mL of HCl 0.1 mol L⁻¹ solution for 2 h. The membrane was recovered; its faces quickly dried out with blotting paper, placed on the moisture analyzer’s balance and the heating process started to reach 120 °C. The measurement stopped when the weight on the balance was changing twice for less than 10⁻³ mg. The final weight of the membrane was considered as the one of the dried membrane m_d. The weight measured right before starting the heating process was considered as the wet weight m_w. The water content T_water was calculated as follows: T_water = 100 × (m_w − m_d)/m_d.

EIS measurements were obtained at room temperature from alternative current impedance analysis using a Solartron 1260 impedance analyzer coupled to a Solartron 1287 potensiostat, over the frequency range 1–10⁷ Hz with controlled voltage. The home-made mercury cell used for the measurements is constituted by two identical compartments, carved in Teflon pieces, between which the membrane is blocked using Viton o-rings [45]. Prior to measurement, the membrane was immersed in a 1 mol L⁻¹ NaOH solution for 24 h and afterwards rinsed and soaked in ultrapure water for 24 h. Right before EIS measurement, the membrane was removed from water (its faces wiped out with blotting paper, its wet thickness measured using a micrometer) and was finally placed in the mercury cell. All the process shall be carried out as swiftly as possible, usually in less than 2 min. Once the membrane was positioned in the cell, the two compartments were filled with mercury; a Pt electrode was dipped in each one and connected to the analyzer. The resistance R of the membrane was derived from the intersection of the high-frequency semicircle on a complex impedance plane with the real Re(Z) axis. The conductivity was calculated using the relationship σ = Th_w/(R × S), where Th_w is the wet thickness of the membrane, S is the active surface of measurement (0.785 cm² here).

NaBH₄ and glycerol permeability were measured using a Hittorf diffusion cell constituted by two carved Teflon compartments compressing the membrane. One compartment contained 20 mL of ultrapure water and the other 200 mL of either a solution composed of a mixture of NaOH (1 mol L⁻¹) and NaBH₄ (2 mol L⁻¹) or a solution of glycerol (2.5 mol L⁻¹). Samples of the smallest compartment solution were taken at regular intervals and analyzed for the amount of NaBH₄ or glycerol present. Boron elemental analysis was carried out by the Service Central d’Analyse of the CNRS (Vernaison, France) and glycerol was titrated by using FT-IR. Measurements were performed in attenuated total reflectance (ATR) mode on either a Nicolet 710 Thermo Nicolet instrument or a Nexus Thermo Electron instrument. Spectra were recorded in transmittance mode between 400 and 4,000 cm⁻¹ wavenumbers, with a resolution of 4 cm⁻¹ and by accumulating 32 scans. Glycerol content was determined by integration of the bands observed in the range 950–1150 cm⁻¹, characteristic of the C–O stretching of primary and secondary alcohols [11,46]. Calibration of the spectrometer was performed using 0.5, 1.0, 2.0 and 2.5 mol L⁻¹ glycerol solutions. Diffusion coefficients were calculated, for a given time of permeation, from the slope α of the Δn = F(ΔC) curves, where Δn represents the molar amount of permeate which has actually permeated and ΔC the permeate concentration gradient between the up- and downstream compartments. The diffusion coefficient value D was calculated from the formula D = α × Th/(Δt × S), where Th is the initial thickness of the membrane, Δt the time of permeation and S the active surface of measurement (0.785 or 3.14 cm² here, depending on the permeation cell used). The flow of permeate J was calculated from the following formula J = (D × ΔC₀)/Th, where ΔC₀ is the initial permeate concentration gradient.

Virgin ADP-Morgan^® and some modified membranes of the same kind were tested in fuel cells. Before integration in membrane-electrodes assemblies (MEA), membranes were activated by immersion in a freshly prepared solution of 1M NaOH for 24 h at room temperature, and then rinsed with ultrapure water and hydrated for one hour. The manufacturing of MEA was performed (at P’ Institute—UPR CNRS 3346) by mechanical pressing of two electrodes against a membrane at ambient temperature. Electrodes were prepared (at LACCO laboratory—UMR CNRS 6503) by deposition of a catalyst ink (Pt (40 wt %)/C, 2 mg_Pt cm⁻²) on a home-made gas diffusion layer [12]. The fuel cell tests were performed (at P’ Institute—UPR CNRS 3346) using a test bench of Fuel Cell Technologies, Inc. including a single cell with an active area of 5 cm². The MEA was placed between two mono-polar plates and assembled under 8 Nm torque wrench [47,48,49,50]. The fuel cell ran with pure oxygen as oxidant at the cathode (100 mL/min, 1 atm. back pressure, 20–80 °C) and a glycerol 1M/NaOH 4M solution as fuel at the anode (10 mL/min, 1 atm. back pressure, 20 °C).