*2.4. Characterizations*

The PEO-based membranes were characterized, both dry and after gelation by the two liquid electrolytes.

Gelation of membranes was realized by dropping the liquid electrolyte on disks of membranes, keeping the weight ratio solution-to-membrane ≈5.

The morphology of the electrospun pristine dry membranes was evaluated by means of scanning electron microscopy (SEM—EVO50, Zeiss, Jena, Germany). In order to reduce the charge accumulation, the membranes were covered with a thin layer of evaporated gold before SEM measurements.

Thermal properties were evaluated by means of differential scanning calorimetry (DSC) and thermal gravimetric analysis (TGA). DSC measurements on both pristine and gelled membranes, as well as on pure PEO powder, were performed with a Mettler-Toledo DSC 821 (Zaventem, Belgium) instrument under an inert nitrogen flux, cooling from room temperature down to −90 ◦C, holding for 10 min at −90 ◦C, and then heating up to 80 ◦C at a scan rate of 10 ◦C/min. TGA was carried out on pristine membrane samples and on pure PEO powder, with a Mettler-Toledo TGA/SDTA 851e under an inert nitrogen flux, heating from room temperature to 600 ◦C at a scan rate of 10 ◦C/min.

Electrochemical impedance spectroscopy (EIS) was used to determine the conductivity of the two swelling solutions and of the gelled membranes. To obtain the conductivity values of the solutions, the measurements were carried out by dipping a conductivity cell for liquids (composed of two sheets of platinum facing at the distance of 1 cm) in the test solution and controlling the temperature with a oven (Büchi-Oven B-585, BUCHI Italia S.r.l., Cornaredo, Italy) in the range 20–60 ◦C. A VSP potentiostat/galvanostat (Bio-Logic Science Instruments, Seyssinet-Pariset, France) was used to record the impedance spectra of the samples in the frequency range of 1 mHz–1 kHz with a sinusoidal signal of 5 mV amplitude. All the spectra, plotted as Nyquist plots, showed an almost vertical straight line intercepting the real axis at high frequency. The value in Ω of this intercept, i.e., the cell resistance at infinite frequency, has been used to calculate the conductivity of the samples under study through the equation:

$$\sigma = 1/\mathbb{R} \times (\mathbb{L}/\mathbb{S})\_{\prime\prime}$$

where "L" and "S" are, respectively, the distance in centimeters and the surface area in centimeters square of the electrodes.

The conductivity of gelled membranes was evaluated using EIS, assembling coin-type cells with stain-less steel current collector electrodes, where the swollen PEO membrane acts as an electrolyte separator. For each cell, a 100 μm-thick Teflon O-ring spacer was adopted and two disks of membrane having diameter 0.8 cm were directly gelled in the cell by dropping the desired amount of liquid electrolyte. Impedance spectra were recorded by applying a 10 mV amplitude signal in the frequency range 200 kHz–1 kHz using a VMP2 potentiostat/galvanostat (Bio-Logic Science Instruments, Seyssinet-Pariset, France). The temperature was controlled in the range −50 ◦C to +60 ◦C with a Tenney Junior Compact Temperature Test Chamber (TPS, White-Deer, PA, USA).

#### **3. Results and Discussion**

Morphology of pristine PEO-based electrospun membranes is shown in Figure 1. SEM images were recorded for both es-PEO (Figure 1a) and es-PEO-SiO2 (Figure 1b) samples, without and with the inorganic additive, respectively. All samples are made of bead-free fibers with an average diameter of 250–300 nm. When nanoparticles were added to PEO solution, the resulting fibers showed slightly higher diameter that can be explained by the inclusion of silica particles inside the polymer fibers. Moreover, Figure 1b shows a certain degree of silica agglomerates, deposited on single fibers or within voids in the polymer mat. Such micrometric agglomerates resulted from the aggregation during the electrospinning process in aqueous media of silica particles, given as spherical porous nanopowder by the product specification.

Thermal stability of pristine membranes, compared to pure PEO powder, was investigated by TGA measurements, as shown in Figure 2.

**Figure 1.** SEM images of the two electrospun membranes: (**a**) without silica additive, and (**b**) with 10 wt.% SiO2.

**Figure 2.** TGA (**a**) and derivative-TGA (**b**) curves of the two electrospun pristine membranes and of pure PEO powder.

High thermal stability typical of pure PEO polymer, extending up to 400 ◦C, is preserved in both pristine membranes. As highlighted by the minimum of the derivative curves in Figure 2b, temperatures of decomposition were quite similar for all the investigated samples (i.e., 401 ◦C for the es-PEO membranes and ≈404 ◦C for both PEO powder and es-PEO-SiO2). Unexpected residual masses were revealed above 450 ◦C in Figure 2a. Pure PEO powder was not completed removed beyond its decomposition temperature, giving ≈8% of left over weight. This could be attributed to non-volatile residuals and to the presence of thermo-stable additives or catalytic compounds used in the synthesis of the polymer. It should be noticed that such residual mass is expected to be reduced by lowering the heating rate, due to a diffusion-limited elimination of the decomposition products by the powder bulk. This did not happen in the case of es-PEO membrane, showing no residuals after it decomposes. Products of decomposition were easily removed from this high surface area sample and eventual additives, such as inhibitors and stabilizers, present in the starting powder were separated and eliminated during the electro-spinning process in aqueous media. A residual mass slightly higher than 10% was revealed for the es-PEO-SiO2 membrane, which was attributed to the inorganic silica filler.

The DSC response of the two pristine membranes, compared with pure PEO powder, is displayed in Figure 3. One main thermal transition was evident in all the DSC traces due to the melting of the polymer crystalline phase. The temperature and energy involved in this melting process have been evaluated and reported in Table 1. Temperature values here shown have been derived from the minimum of the endothermic peak, and in this respect, no valuable differences are noticed among the samples. The additive-free electrospun membrane is actually the sample showing a slightly lower melting temperature and a narrower peak. More remarkable differences were observed in terms of the enthalpy change, highlighting the role of both the electrospinning process and the silica filler. Clearly, lower energy was involved in the melting of pure PEO powder, which also revealed its lower crystallinity. The electrospinning process, due to its ordering effect, somehow increased the crystalline degree of the membranes, compared to the starting polymer powder, corresponding to a higher melting enthalpy. On the contrary, silica nanoparticles and their micrometric aggregates have the effect of lowering crystallinity, towards a more amorphous system with respect to plain, PEO-based electrospun membranes.

**Figure 3.** Heating scan of the DSC curves recorded for the two electrospun pristine membranes and for pure PEO powder.

**Table 1.** Enthalpy change and temperature related to PEO melting (values derived from DSC response in Figure 3).


Gelation of the polymer membranes using liquid electrolyte solutions (i.e., PS-free sol and PS-containing sol) was achieved to finally obtain the desired Li+-conducting composite electrolytes. The transition from solid to gel-like systems was very easily and quickly attained with the proposed electrospun fiber mats due to their high surface-to-volume ratio. Thermal properties of the resulting electrolytes have been checked after the gelation process. The DSC response of the new gelled electrolytes is reported in Figure 4. As expected, the melting transition of the polymer was highly influenced by the presence of the liquid component. The crystallinity of PEO was strongly reduced, almost suppressed, upon gelation, giving rise to an amorphous, plasticized electrolyte system. In this respect, the nature of the electrolyte solution, with or without polysulfide, appeared almost irrelevant. The main role was due to the ionic liquid (i.e., the major component of the liquid electrolytes), which interacted with the polymer chains, thus preventing their crystallization. Interestingly, a certain degree of crystalline phase, even though very small, was preserved when silica particles were present (see the melting peak around 60 ◦C in Figure 4b). This is quite reasonably attributed to preferential interactions established between the inorganic filler and the liquid solution, leaving partially-coordinated PEO chains free to crystallize. In this respect, if we assume the starting PEO powder as a reference, it was possible to compare the crystallinity of our different samples by dividing the enthalpy change for the melting transition of each quoted sample by the enthalpy of melting related to the PEO powder. As already pointed out, crystallinity of the electrospun starting membranes was higher than that of the PEO powder due to the ordering effect of the electrospinning process [3], i.e., 1.81 times higher for the es-PEO membrane and 1.44 times higher for the es-PEO-SiO2 membrane. Such an estimate was not possible for silica-free gelled systems, as no melting was observed in the DSC traces. Whereas, a very low crystalline degree was maintained in the silica-added gel polymer electrolytes, with the crystallinity being 0.12 and 0.08 times that of PEO powder for es-PEO-SiO2/PS-free sol and es-PEO-SiO2/PS-containing sol, respectively. Another thermal response was noticed around −80 ◦C in the DSC traces of Figure 4 due to the glass transition of the ionic liquid component [22].

**Figure 4.** Heating scan of the DSC response of additive free PEO membrane (**a**) and of silica-added PEO membrane (**b**) gelled using liquid electrolyte "PS-free sol" (black curves) or "PS-containing sol" (red curves).

Functionality of the gelled membranes as electrolyte was tested using EIS measurements performed at increasing temperature in the range −50 ◦C to 60 ◦C. For comparison purposes, impedance spectra were recorded for the liquid electrolyte solutions as well, in this case limiting the temperature range between 20 ◦C and 70 ◦C. Conductivity values, obtained from the impedance spectra in the investigated T-ranges, are reported in Figure 5a,b for the swelling solutions and for the swollen membranes, respectively, in the form of Arrhenius plots. Conductivity values, extrapolated from the plots in Figure 5b at two temperatures of interest (i.e., 25 ◦C and 50 ◦C, representing normal operating conditions and melting region of the polymer, respectively) are reported in Table 2.

**Figure 5.** Arrhenius plots of conductivity of the swelling liquid solution (**a**) and of the gelled electrolyte membranes (**b**).



With the exception of the es-PEO-SiO2/PS-free sol sample, very high conductivities were achieved for all the gelled systems at both investigated temperatures. It is worth noticing that these room-temperature σ values are typical of viscous organic electrolytes used in lithium-batteries, revealing that the polymer membranes here were very well plasticized. As shown below, a very limited conductivity decrease was observed when moving from the liquid electrolytes, PS-free sol and PS-containing sol, to the gelled polymer systems. As expected, a temperature-activated transport mechanism was found, giving higher conductivity values at 50 ◦C with respect to 25 ◦C.

In Figure 5a, it is possible to observe both the liquid electrolytes showing interesting ionic conductivity, and the presence of polysulfide ions (in PS-containing solution) seemed to affect the conduction properties very little. Overall, the detected conductivity values appear suitable for battery applications in a wide temperature range. The addition of polysulfide in solution has, in general, two opposite effects on the overall ionic conductivity: on one hand, the number of charged species increased due to the intake of anions in the solution; on the other hand, an increase in viscosity is expected, which hindered the ion mobility. In our systems, the two solutions had very high, similar conductivity values, indicating that these effects offset each other.

As shown in Figure 5b, conductivity was lower for the silica-added gel polymer electrolytes (black and green plots) compared to the silica-free systems (blue and red plots). This can be explained by considering a possible retention of the liquid electrolyte fraction on the silica particles, hindering the polymer chain-assisted ion transport. In this respect, a certain role is played by the nature of the liquid electrolyte. Indeed, big differences were observed among the two silica-added membranes according to the type of swelling liquid electrolyte, with the polysulfide-doped solution giving higher overall conductivity with respect to the polysulfide-free system (compare es-PEO-SiO2/PS-containing sol with es-PEO-SiO2/PS-free sol in Figure 5b).

Each curve of Figure 5b was fitted using the Vogel–Tamman–Fulcher (VTF) equation:

$$
\sigma(T) = A \cdot T^{-\frac{1}{2}} \cdot e^{-\frac{E\_A}{R(T-T\_0)}}
$$

with *A* and *EA* parameters representing the charge carrier number and the activation energy, respectively, whereas *T*0 is the ideal glass transition temperature. Very high correlation (higher than 0.99) was found between the experimental and fitted curves, revealing that our systems followed the typical behavior of ion-conducting amorphous matrices where the polymer component assisted the ion transport. The extrapolated parameters were considered and reported in Table 3.

**Table 3.** Parameters of the VTF equation derived by fitting the curves in Figure 5b.


EA values were low, compared to other IL-added PEO-based ion-conducting systems [23], meaning that ion transport and conduction mechanism were very easily activated. Moreover, no substantial differences were found among samples in terms of activation energy. Similarly, comparably low values of T0 were obtained, proving that all the gelled electrolytes exhibited amorphous behavior. Differences of relevance were found in terms of the A parameter. The smallest value was obtained for es-PEO-SiO2/PS-free sol sample, revealing that its low conductivity (see black curve in Figure 5b and values in Table 2) was actually due to a reduced number of total charge carriers. This supports our hypothesis that silica particles absorbed the liquid electrolyte, thus limiting the concentration of ions available for transport. Based on the higher conductivity and A values observed in es-PEO-SiO2/PS-containing sol, we conclude that the polysulfide opposes this retention effect of SiO2 additive to the advantage of the transport mechanism.
