Enhancement of the Green H2 Production by Using TiO2 Composite Polybenzimidazole Membranes

This study reports the hydrogen production using TiO2 based composite polybenzimidazole membranes through the SO2 depolarized electrolysis that requires lower energy input than the direct water electrolysis. Composite membranes prepared and studied in this work showed very promising results in terms of proton conductivity, chemical stability, and crossover. Thus, a reduction in SO2 crossover was observed with the increase of the concentration of TiO2, obtaining reductions as high as 42% with the 3.0 wt% TiO2-PBI membrane at 120 °C. Higher hydrogen production rates and Faradaic efficiencies were achieved by all the composite membranes, with an optimum for the 1.0 wt% TiO2-PBI membrane (with this membrane, the production of hydrogen increased a 53% at 110 °C and a 49% at 120 °C as compared with the standard PBI membrane), demonstrated the benefit of the use of composite membranes with respect to the standard one for green hydrogen production.


Introduction
Renewable energies are of paramount importance for a clean and sustainable development of the energy sector. However, the transition from a carbon-based sector to a green energy sector has some steps that are of concern and challenges that have to be addressed. One of them is that renewable energy production is intermittent and very difficult to predict. Therefore, the energy sector cannot rely on them to meet the energy demand of the population. There can be availability problems of electricity produced by these renewable energies that would cause shortage [1]. Consequently, the development of energy storage technologies capable to store the excess of produced renewable energy and to release it when needed is key to address these challenges. A promising method to store these energies is to do it in the form of hydrogen as energy vector [2]. For green hydrogen production, water electrolysis is the benchmark. However, due to the high energy requirements for this process, new alternatives are being developed [3][4][5][6]. Among those alternatives, the Hybrid Sulfur (HyS) cycle is an increasingly popular process for this green hydrogen production [7]. The HyS cycle has three steps of which the main one is the electrolysis step, in which hydrogen is generated, which benefits of a very low E 0 = 0.158 V compared to the E 0 = 1.210 V of water electrolysis. Therefore, the first of the three steps is the electrolysis (sulfur dioxide depolarized electrolysis, SDE) of water and sulfur dioxide (Equation (1)), which produce protons that recombine into hydrogen (Equation (2)) in the cathode of the electrolyzer and sulfuric acid which, in the second step, is thermally decomposed into water, sulfur dioxide and oxygen (Equation (3)). Finally, in the third step, oxygen and sulfur dioxide are separated, and then SO 2 is recirculated to the electrolysis step.
2H + +2e − → H 2 (2) elsewhere [28]. Briefly, TiO 2 nanoparticles were dispersed in DMAc for 15 min in an ultrasound bath. Then, in a different vial, a diluted PBI solution was prepared by adding DMAc to the commercial solution. Once a homogeneous PBI solution was obtained, both solutions were mixed, obtaining a final PBI concentration of 2.0 wt% for membrane casting. This solution was further homogenized in the ultrasound bath for 2 h. The films were obtained by pouring the prepared solutions in a glass plate (13 cm diameter) and evaporating the solvent at 80 • C in an oven for 24 h. The final membranes were detached form the plate using DI water [34,35]. Prior to doping the membranes for further characterization and for the electrolysis experiments, they were washed with boiling water during 2 h and then dried again at 80 • C.

Chemical and Physicochemical Characterization 2.3.1. XRD & SEM
XRD spectra for the pristine and composite membranes was obtained by X-Ray Diffraction in a Philips X'Pert MPD (PANALYTICAL, Malvern, UK) diffractometer applying Kα corresponding to the transition from copper radiation (λ = 1.5404 Å). Data were obtained recording the 2θ angular region from 10 • to 100 • (scan rate 0.02 • ·s −1 ), and 2 cm 2 samples were used.
The SEM pictures were carried out in a Microscope Gemini SEM 500 (Oberkochen, Germany) field emission. To prepare the samples, all the membranes were sputter-coated with a very thin layer of gold. The thickness of this conductor layer was 2 nm.

Ionic Conductivity
The conductivity of the membranes was measured using the four-points system. The system is formed by a stainless-steel plate (0.1 cm of thickness) with four perforations where gold wires are placed. The plate was electrically insulated using a Teflon sheet (W.L. Gore and Associates, Newark, DE, USA). The sample (6.0 × 1.0 cm) is placed on the top of the plate and the wires are placed on the top of the membrane which are connected to a galvanostat/potentiostat (AutoLab PGSTAT204, Barendrecht, The Netherlands) equipped with a frequency response analysis module (FRA32 module). The device was thermally insulated from the exterior while performing the measurements and connected to a heater. After 1 h of reaching the desired temperature, electrochemical impedance spectroscopy (EIS) analysis was carried out (frequency range 100-10,000 Hz and amplitude of 10 mV). Proton conductivity was calculated using Equation (4). R Ω (ohm) represents the obtained value of the resistance to the ionic flux, S (cm 2 ) is the transversal section of the sample, and l (cm) is the distance between the wires where a potential difference is measured.

Acid Doping Level (ADL)
Samples of 2.0 × 2.0 cm 2 were soaked in 85 wt%H 3 PO 4 . The moisture of the membranes was first removed by drying them at 80 • C overnight (m dry PBI ). Then, the membranes were weighted and doped in phosphoric acid for 24 h at 60 • C. When the wet weight of the doped PBI membranes was constant, the samples were dried again to remove the absorbed water. Then, the membranes were weighed (m doped PBI ) again to calculate the acid uptake using Equation (5

Phosphoric Acid Retention Capability
Acid retention of the prepared membranes was measured as follows. Doped samples (2.0 × 2.0 cm 2 ) were immersed in a glass bottle with deionized water at 80 • C for one hour while magnetically stirred. The concertation of phosphates was monitored during the experiment by taking liquid samples each 10 min, which were measured by ionic chromatography.

Contact Angle Measurement
Water management is a very important parameter for water splitting processes. Therefore, the contact angle of a drop of water was used to determine the affinity of the membranes towards water. To do so, a water drop is dropped on top of the sample and a picture is taken after 15 min of stabilization. Then, the right and left contact angles are determined from the interface of the water drop and the membrane.

Chemical Stability
Chemical resistance, of the nanocomposite membranes and the pristine PBI membrane, was measured by performing an accelerated oxidation test with the sulfate radical as oxidizing agent (SO 4 − ·). The selection of this radical (and not OH·) was due to the very high oxidation power of this radical [36] which is more likely to be formed in the sulfur based electrochemical reactor due to side reactions at high [37] in the acidic sulfur environment. For the persulfate test, dried PBI based membranes (45 mg) were introduced in a flask with an initial concentration of Na 2 S 2 O 8 (Sigma Aldrich, Burlington, MA, USA) of 500 ppm in H 2 SO 4 1.0 M. The experiments were carried out at 80 • C, until membrane failure was observed.

SO 2 Crossover Measurement
Sulfur dioxide crossover was measured by means of electrochemistry as follows. MEASs were prepared with a catalyst loading of 0.7 mgPt·cm −2 in both anode an cathode with an membrane area of 4 cm 2 . The working electrode was fed with a mixture of steam and nitrogen. The counter electrode was first fed with nitrogen and a constant voltage of 0.7 V was applied to create a baseline and once it was obtained, SO 2 (70 mL·min −1 . Then, the new current value obtained is recorded until reaching a constant value. The crossover was calculated as the gap between the baseline and the current when flowing sulfur dioxide that crossed the membrane and underwent oxidation according to Equation (1). This process was repeated for all the studied temperatures. Thus, the flux of SO 2 which crosses the membrane is proportional to the net measured current. This procedure was also followed by Opperman et al. to measure the transport of SO 2 through PBI membranes [38].

Sulfur Dioxide Depolarized Electroylsis
A 25 cm 2 electrolyzer was employed for the electrochemical characterization for the pristine and composite membranes ( Figure S1 depicts the experimental set-up). In order to evaluate the effect of temperature, the electrochemical cell was characterized in the temperature range of 110 • C and 140 • C. Prior the experiments, the membranes were doped in 85 wt% H 3 PO 4 at 80 • C for 12 h. Both anode and cathode were prepared with a catalyst loading of 0.7 mgPt·cm −2 by air dispersion of an ink containing commercial Pt/C (40 wt% of Pt with Vulcan XR-72) with PBI as ionomer and DMAc as solvent. This ink was sprayed onto H23C2 GDL gas diffusion layers (Freudenberg, Weinheim, Germany). The electrolyzer was finally assembled with eight equidistant bolts and employing a compression toque of 1 N-m.
Electrolyzer operation was carried out as follows. Water flow was 0.2 mL min −1 in liquid phase (λ = 2) and SO 2 gaseous flow was 70 mL·min −1 in the anode of the electrolyzer. Before entering the electrolyzer, water and SO 2 were mixed in a "T" connection. Then, they pass through a stainless-steel tube connected to a heater (110 • C) to vaporize the water. On the other hand, a mixture of nitrogen (100 mL min −1 ) and steam (0.5 mL min −1 of liquid water) is fed. For each temperature, the anode was collected to measure the production of H 2 SO 4 at 0.60 V by measuring the produced sulphates by ionic chromatography. The system was electrochemically characterized using of linear sweep voltammetries (LSVs) (0.1 to 1.0 V at a scan rate of 10 mV s −1 ) using a galvanostat/potentiostat (AutoLab PGSTAT204, Barendrecht, Netherlands). EIS measurements were carried out at 0.7 V with an amplitude of 20 mV and a frequency range of 0.1-10,000 Hz. A model R(RC), shown in Figure S2, was employed for the fitting of the experimental EIS data to obtain the ohmic and charge transfer resistance values. Hydrogen obtained in the cathode was characterized by gas chromatography with a GC-2030 (Shimadzu, Japan) equipped with a Rxi-1ms column (L = 30 m; ID = 0.32 mm; DF = 0.50 µm) for sulfurous compounds (H 2 S and SO 2 ) and a Rt-Msieve 5A column (L = 30 m; ID = 0.32 mm; DF = 30 µm) for the identification of small gas molecules (H 2 , N 2 and O 2 ). Figure 1 shows the XRD data of the PBI composite membranes with different TiO 2 contents and in the inset the XRD pattern of the TiO 2 used as filler. It can be observed that the main peaks of the TiO 2 compound corresponds with the crystalline phase of Rutile as the mean peaks of rutile phase at 2Θ =27.5 • , 54.3 • , and 36.1 • appear [39]. On the other hand, the typical broad peak at 2Θ =25 • , which results from the convolution of amorphous and crystalline scattering, is observed [31]. Regarding the composite PBI based membranes, it can be clearly distinguished that the higher the content of rutile was, the more intense was the main peak. Furthermore, the other peaks of the TiO 2 rutile phase appeared more clearly in the plot when the content of TiO 2 was 2.0 wt% or higher. These results confirm that the inorganic filler is within the polymer matrix and the structure of it has not changed when it was introduced in the polymeric film.  On the other hand, SEM surface characterization ( Figure 2) shows a good interaction between the TiO2 nanoparticles and PBI. As observed, the surfaces do not show imperfections apart from the apparition of TiO2 particles covered by the polymer, which indicates good compatibility between PBI and the inorganic filler.

Results and Discussion
Two key parameters for PEM electrolyzers are the water management and the chemical stability of the membranes. Regarding water management, the contact angle of the membranes was measured ( Figure S3). When a small amount of TiO2 is added to the PBI matrix, the behaviour of the membrane changes from hydrophilic to an almost hydropho-  On the other hand, SEM surface characterization ( Figure 2) shows a good interaction between the TiO 2 nanoparticles and PBI. As observed, the surfaces do not show imperfections apart from the apparition of TiO 2 particles covered by the polymer, which indicates good compatibility between PBI and the inorganic filler. The doping level and acid retention capability are two main features of phosphoric acid doped PBI based membranes to be evaluated for SO2 depolarized electrolysis for hydrogen production as the first one is related to the conductivity and the other to the durability issues of the system. In Table 1, it can be seen that these two parameters and the thickness, as well, increase when the membrane is doped for all the membranes studied in this work. At a first sight, it could be said that there is not a clear tendency regarding the doping level with the content of TiO2. Nevertheless, it can be stated that these results agree with previous results when different contents (2.0 wt%, 4.0 wt%, 8.0 wt% and 16.0 wt%) of TiO2 were added to a PBI membrane for a HT-PEMFC [30]. In that study, the optimum filler content was 2.0 wt% as higher values of TiO2 led to doping levels even lower than those attained in the pristine PBI membrane. It has been found in the literature, that inorganic oxides tend toward the agglomeration in the polymeric solution and, after solvent evaporation step, the active surface for acid absorption diminished [40,41]. In this work, the range of TiO2 content is narrower as we knew those results and, in this case, it seems that small amounts of TiO2 as filler are adequate to improve the doping level with respect to the standard PBI membrane, being 0.5 wt% the one with the highest doping level and, hence, the one with the highest thickness increase (as this parameter is closely related with the doping level). Thus, for the case of the best membrane, from the doping Two key parameters for PEM electrolyzers are the water management and the chemical stability of the membranes. Regarding water management, the contact angle of the membranes was measured ( Figure S3). When a small amount of TiO 2 is added to the PBI matrix, the behaviour of the membrane changes from hydrophilic to an almost hydrophobic performance. This can be attributed to changes in the surface of the membrane associated to the introduced particles. In addition, the similar values for the right and left contact angle indicate a homogeneous surface of the membranes. Chemical stability of the prepared membranes was evaluated by means of sulfate radicals attack ( Figure S4), which are expected to be formed in a sulfur environment. There is a great improvement when adding TiO 2 as the time at which the membrane broke increased from 8 h for the pristine PBI membrane to 30 h for the 0.5 wt% TiO 2 -PBI membrane, 40 h for the 1.0 wt% TiO 2 -PBI membrane, 42 h for the 2.0 wt% TiO 2 -PBI membrane and 42 h for the 3.0 wt% TiO 2 -PBI membrane. Thus, adding a filler created a more compact structure capable of protecting the polymer chains from the attack of the sulfate radicals. These results are in line with the addition of other fillers to the PBI matrix which also increase it chemical stability [28].
The doping level and acid retention capability are two main features of phosphoric acid doped PBI based membranes to be evaluated for SO 2 depolarized electrolysis for hydrogen production as the first one is related to the conductivity and the other to the durability Nanomaterials 2022, 12, 2920 7 of 15 issues of the system. In Table 1, it can be seen that these two parameters and the thickness, as well, increase when the membrane is doped for all the membranes studied in this work. At a first sight, it could be said that there is not a clear tendency regarding the doping level with the content of TiO 2 . Nevertheless, it can be stated that these results agree with previous results when different contents (2.0 wt%, 4.0 wt%, 8.0 wt% and 16.0 wt%) of TiO 2 were added to a PBI membrane for a HT-PEMFC [30]. In that study, the optimum filler content was 2.0 wt% as higher values of TiO 2 led to doping levels even lower than those attained in the pristine PBI membrane. It has been found in the literature, that inorganic oxides tend toward the agglomeration in the polymeric solution and, after solvent evaporation step, the active surface for acid absorption diminished [40,41]. In this work, the range of TiO 2 content is narrower as we knew those results and, in this case, it seems that small amounts of TiO 2 as filler are adequate to improve the doping level with respect to the standard PBI membrane, being 0.5 wt% the one with the highest doping level and, hence, the one with the highest thickness increase (as this parameter is closely related with the doping level). Thus, for the case of the best membrane, from the doping point of view (the one with 0.5 wt%), the doping level increased 11.4% and the thickness 13.8% with respect to the standard PBI. Acid leaching tests for all the membranes were performed under the same operation conditions to assess the acid retention capacity, which is considered one of the main causes of a loss of conductivity for fuel cell technology and, hence, for electrolysis applications [42][43][44]. It can be observed in Table 1, the positive effect of the presence of TiO 2 as filler, regardless of the content, at least, the acid retention capacity doubled with respect to the pristine PBI membrane. Furthermore, the acid retention capability increases more than three times when the content of titania is 1.0 wt%, and this effect is achieved also by the membranes with 2.0 and 3.0 wt%. The effect observed agrees with our previous work where TiO 2 and TiOSO 4 were used as fillers in PBI based membranes [45], and with results obtained by Mustarelli et al. [43] where it was found that the addition of 5.0 wt% of silica in a PBI matrix reduced the acid leaching from the membrane by more than a factor of two.
Proton conductivity is one of the most useful parameters to anticipate the behaviour of the membrane in an electrolyzer. In-plane conductivity of the standard and the composite TiO 2 -PBI doped with phosphoric acid is shown in Figure 3. It can be observed, that the in-plane conductivity is higher for all the composite membranes compared with the pristine PBI membrane, with a maximum in conductivity for all the samples at 130 • C. Temperature does not play a significant role in this case, because there are not significant changes when increasing temperature in the studied range. On the other hand, the slight decrease in conductivity when temperature is higher than 130 • C can be related to phosphoric acid demineralization associated to losses of water, as has been pointed out by Ma et al. [46]. On the other hand, a considerable increase is obtained for the composite membrane with a TiO 2 content of 0.5 wt% with a value of 0.086 S cm −1 at 130 • C as compared with the conductivity of 0.058 S cm −1 , measured for the standard PBI membrane. Then, the conductivity of the composite membranes decreases when the amount of the filler increases. This behaviour can be correlated with the acid doping level ( Table 1). The increase in acid doping level with a maximum for the 0.5 wt% TiO 2 membrane explains its conductivity above the other membranes, then the in-plane conductivity decreases with the decrease in the ADL.
Nevertheless, the conductivity for the 1.0 wt% TiO 2 composite membrane is 28% higher than for the pristine membrane (0.074 S cm −1 ), 12% higher for the 2.0 wt% composite membrane (0.065 S cm −1 ), and 5.0% higher for the 3.0 wt% TiO 2 membrane. The electrolysis performance for the prepared membranes is shown in Figure 4. The addition of TiO2 to the PBI matrix clearly benefits the electrolyzer performance at 110 °C and 120 °C for a filler concentration of 1.0 wt%. For example, a current density of 0.13 A cm −2 at 0.6 V and 0.34 A cm −2 at 0.85 V was obtained for the 1.0 wt% TiO2-PBI membrane which was a value 44% higher at 0.6 V and 21% higher at 0.85 V than for the pristine PBI membrane (0.09 A cm −2 and 0.28 A cm −2 respectively). Then, at 120 °C, there is a slight improvement at a potential of 0.6 V as the 1.0 wt% membrane reaches a current density of 0.15 A cm −2 , demonstrating that temperature enhances the kinetics in the electrolyzer. However, for the standard membrane, a current density of 0.07 A cm −2 at 0.6 V is obtained, increasing the improvement of a TiO2 content of 1.0 wt% to 114% in terms of current density compared to the pristine PBI membrane. Increasing the filler content to a 2.0 wt% also improves the results of the pristine PBI membrane, however, this membrane showed current density values lower than the 1.0 wt% TiO2-PBI membrane. This can be attributed to the in-plane conductivity, which is lower when increasing the filler content. On the other hand, for the minimum and maximum TiO2 contents, the obtained results worsen those obtained for the pristine membrane. In the case of the 0.5 wt% TiO2 membrane, the high conductivity obtained can be rapidly lost due to the lower acid retention of this membrane and, also, the low content of TiO2 in this membrane might not be enough to see any improvement in the electrolyzer. Then, the membrane with the highest content of TiO2 presented a very similar conductivity as compared with the standard membrane. Moreover, a high filler content can disrupt proton transport channels through the membrane, decreasing electrolyzer performance [45]. At 130 °C and 140 °C, there is a noticeable loss of performance in the electrolyzer. Figures 4c,d show almost no difference in the polarization curves when the electrolyzers are equipped with the pristine PBI membrane or any of the TiO2 composite membranes. This effect of the temperature is attributed to side reactions that can happen in the HyS process due to SO2 crossover [47][48][49]. These results exhibit that side reactions (Equations (7)- (9)) are favoured at temperatures higher than 130 °C. Therefore, the catalyst can be poisoned by the formation of elemental sulfur (Equations (7) and (9)) or by the formation of hydrogen sulfide (Equation (8) The electrolysis performance for the prepared membranes is shown in Figure 4. The addition of TiO 2 to the PBI matrix clearly benefits the electrolyzer performance at 110 • C and 120 • C for a filler concentration of 1.0 wt%. For example, a current density of 0.13 A cm −2 at 0.6 V and 0.34 A cm −2 at 0.85 V was obtained for the 1.0 wt% TiO 2 -PBI membrane which was a value 44% higher at 0.6 V and 21% higher at 0.85 V than for the pristine PBI membrane (0.09 A cm −2 and 0.28 A cm −2 respectively). Then, at 120 • C, there is a slight improvement at a potential of 0.6 V as the 1.0 wt% membrane reaches a current density of 0.15 A cm −2 , demonstrating that temperature enhances the kinetics in the electrolyzer. However, for the standard membrane, a current density of 0.07 A cm −2 at 0.6 V is obtained, increasing the improvement of a TiO 2 content of 1.0 wt% to 114% in terms of current density compared to the pristine PBI membrane. Increasing the filler content to a 2.0 wt% also improves the results of the pristine PBI membrane, however, this membrane showed current density values lower than the 1.0 wt% TiO 2 -PBI membrane. This can be attributed to the in-plane conductivity, which is lower when increasing the filler content. On the other hand, for the minimum and maximum TiO 2 contents, the obtained results worsen those obtained for the pristine membrane. In the case of the 0.5 wt% TiO 2 membrane, the high conductivity obtained can be rapidly lost due to the lower acid retention of this membrane and, also, the low content of TiO 2 in this membrane might not be enough to see any improvement in the electrolyzer. Then, the membrane with the highest content of TiO 2 presented a very similar conductivity as compared with the standard membrane. Moreover, a high filler content can disrupt proton transport channels through the membrane, decreasing electrolyzer performance [45]. At 130 • C and 140 • C, there is a noticeable loss of performance in the electrolyzer. Figure 4c,d show almost no difference in the polarization curves when the electrolyzers are equipped with the pristine PBI membrane or any of the TiO 2 composite membranes. This effect of the temperature is attributed to side reactions that can happen in the HyS process due to SO 2 crossover [47][48][49]. These results exhibit that side reactions (Equations (7)- (9)) are favoured at temperatures higher than 130 • C. Therefore, the catalyst can be poisoned by the formation of elemental sulfur (Equations (7) and (9)) or by the formation of hydrogen sulfide (Equation (8)). Considering the obtained results, the depolarized electrolysis of SO2 is not the only sulfur-based reaction taking place in the reactor and a deep product characterization should be carried out to observe the formation of those un-desired products. The drastic change in the behaviour of the electrolyzer when increasing temperature is also observed in the EIS data shown in Table 2 (Nyquist plots reported in Figure S5). One of the parameters that EIS allows to determine is the ohmic resistance of the reactor, with the membrane being the main contributor for this resistance. Variations in the ohmic resistance are attributed to the different composition of the membranes. The lowest value is obtained for the composite membrane 1.0 wt% TiO2-PBI, which already showed a high in-plane conductivity and a high acid retention. The higher value in the resistance measured for the membrane with a TiO2 content of 0.5 wt% can be explained by a low acid retention which, in turn, causes an increase in resistance over the course of the experiment. It can be observed that temperature does not play a relevant role for the ohmic resistance, in fact, the formation of elemental sulfur (Equations (7) and (9)) could minimize the beneficial effect of temperature [14,28]. Regarding the charge transfer resistance, the obtained results help to understand the trend of the linear sweep voltammetries reported in Figure  4. When adding TiO2 to the PBI matrix, a positive effect is observed in all the range of temperature. However, only for contents of 1.0 wt% and 2.0 wt% this improvement is of Considering the obtained results, the depolarized electrolysis of SO 2 is not the only sulfur-based reaction taking place in the reactor and a deep product characterization should be carried out to observe the formation of those un-desired products.
The drastic change in the behaviour of the electrolyzer when increasing temperature is also observed in the EIS data shown in Table 2 (Nyquist plots reported in Figure S5). One of the parameters that EIS allows to determine is the ohmic resistance of the reactor, with the membrane being the main contributor for this resistance. Variations in the ohmic resistance are attributed to the different composition of the membranes. The lowest value is obtained for the composite membrane 1.0 wt% TiO 2 -PBI, which already showed a high in-plane conductivity and a high acid retention. The higher value in the resistance measured for the membrane with a TiO 2 content of 0.5 wt% can be explained by a low acid retention which, in turn, causes an increase in resistance over the course of the experiment. It can be observed that temperature does not play a relevant role for the ohmic resistance, in fact, the formation of elemental sulfur (Equations (7) and (9)) could minimize the beneficial effect of temperature [14,28]. Regarding the charge transfer resistance, the obtained results help to understand the trend of the linear sweep voltammetries reported in Figure 4. When adding TiO 2 to the PBI matrix, a positive effect is observed in all the range of temperature. However, only for contents of 1.0 wt% and 2.0 wt% this improvement is of relevance as for the other two composite membranes the charge transfer resistance is very similar to the obtained for the pristine PBI membrane. The lower acid doping level of the 3.0 wt% TiO 2 -PBI membrane and its very similar conductivity to the pristine PBI membrane explains the results shown in Table 3. Then, as aforementioned, the low performance of the 0.5 wt% TiO 2 -PBi membrane is attributed to its low phosphoric acid retention. When increasing temperature from 120 • C to 130 • C, the charge transfer resistance drastically increases, demonstrating that side reactions are being favoured at those temperatures. Even though obtaining high current densities is of importance for the operation of the electrolyzer, as it gives an idea of the reaction rate, the most essential parameter is the production of hydrogen, as its obtention is the purpose of the water molecule splitting processes. However, this production of hydrogen has never been reported for other HyS works, except for our previous work [28]. Figure 5 shows the actual production of hydrogen as function of temperature and the Faradaic efficiency for each of the employed membranes. Faradaic efficiency was calculated using Equation (10), where I is the intensity recorded at 0.6 V, t is the unit of time, F is faraday's constant (96,500 C mol −1 ), n the number of electrons (2), and ρ H 2 is the hydrogen density (0.041 mol L −1 ) at 20 • C and 1 bar.

Faradaic efficiency [%] =
Actual H 2 flow L·s −1 The benefit of the incorporation of TiO 2 as inorganic filler is clearly observed at 110 • C, obtaining higher hydrogen production rates and Faradaic efficiencies for the composite membranes, with an optimum for the 1.0 wt% TiO 2 -PBI membrane. For instance, 15.23 mL min −1 of hydrogen are obtained when the electrolyzer is equipped with the 1.0 wt% TiO 2 -PBI membrane and 13.48 mL min −1 for the 2.0 wt% TiO 2 -PBI compared to 9.93 mL min −1 for the pristine membrane. At a higher temperature of 120 • C, hydrogen production and Faradaic efficiency follow the same trend, with higher hydrogen production values for the membranes with a filler content of 1.0 wt% and 2.0 wt%, demonstrating the benefits of TiO 2 as filler for PBI membranes for the HyS cycle. On the other hand, as previously observed in the polarization curves (Figure 4), there is a drastic drop at 130 • C and 140 • C. The hydrogen production rate is much lower than for the previous temperatures and the efficiency towards hydrogen production is also lower. Again, these results agree with the charge transfer resistance data, that prove the side reactions are occurring in the electrolyzer which are consuming the produced hydrogen to produce hydrogen sulfide (Equation (9)) or protons for Equation (8), where elemental sulfur is formed. This hypothesis is further confirmed by the levels of hydrogen sulfide measured from the outlet of the electrolyzer and showed in Figure S6. It can be observed that the production of HyS drastically increases at 130 • C when the production of hydrogen decays. At 140 • C, the formation of elemental sulfur is favoured over H 2 S (Figure S6), as the cathode outlet acquired a yellow colour at this temperature. That is the reason the molar ratio H 2 /H 2 S slightly increases at 140 • C. Furthermore, H 2 SO 4 efficiency was measured by ionic chromatography ( Figure S7). As for hydrogen, the efficiency towards the production of sulfuric acid drops at 130 • C for all the cases, demonstrating that the desired reaction is being limited.
Nanomaterials 2022, 12, x FOR PEER REVIEW 12 of 16 S7). As for hydrogen, the efficiency towards the production of sulfuric acid drops at 130 °C for all the cases, demonstrating that the desired reaction is being limited.  Table 3 reports the molar ratio of H2 and H2S at each temperature for the pristine and the composite membranes. As observed, adding TiO2 as inorganic filler to PBI definitely benefits the electrolyzer performance towards the production of hydrogen. In this case, the composite membrane 1.0 wt% TiO2-PBI outperforms the pristine PBI membrane at 110 °C and 120 °C, and the other composite membranes also show very good results when compared with the pristine PBI membrane. By looking at Table 3, it can also be observed how the formation of hydrogen is inhibited by the formation of H2S when increasing temperature. c. d.
a.  Table 3 reports the molar ratio of H 2 and H 2 S at each temperature for the pristine and the composite membranes. As observed, adding TiO 2 as inorganic filler to PBI definitely benefits the electrolyzer performance towards the production of hydrogen. In this case, the composite membrane 1.0 wt% TiO 2 -PBI outperforms the pristine PBI membrane at 110 • C and 120 • C, and the other composite membranes also show very good results when compared with the pristine PBI membrane. By looking at Table 3, it can also be observed how the formation of hydrogen is inhibited by the formation of H 2 S when increasing temperature.
As aforementioned, the side reactions occurring in the cathode of the electrolyzer are caused by the crossover of sulfur dioxide. The improvement of the composite membranes can be, in part, attributed to a reduction in the crossover of sulfur dioxide. Figure 6 depicts the crossover of SO 2 measured for the prepared membranes. Adding a small amount of 0.5 wt% of TiO 2 already results in a clear reduction of the crossover. The crossover further decreases when increasing the content of the filler and reducing the measured current density due to the crossover of SO 2 from 0.44 mA cm −2 to 0.15 mA cm −2 for the 3.0 wt% TiO 2 -PBI membrane. The addition of the fillers changes the inner structure of the membrane, reducing the free volume in the membrane and, also, acting as barriers for the transport of sulfur dioxide from the anode to the cathode of the cell. These crossover results help to comprehend the hydrogen production results and the best performance of the composite membranes compared to the pristine PBI membrane shown in Figures 4 and 5. As aforementioned, the side reactions occurring in the cathode of the electrolyzer are caused by the crossover of sulfur dioxide. The improvement of the composite membranes can be, in part, attributed to a reduction in the crossover of sulfur dioxide. Figure 6 depicts the crossover of SO2 measured for the prepared membranes. Adding a small amount of 0.5 wt% of TiO2 already results in a clear reduction of the crossover. The crossover further decreases when increasing the content of the filler and reducing the measured current density due to the crossover of SO2 from 0.44 mA cm −2 to 0.15 mA cm −2 for the 3.0 wt% TiO2-PBI membrane. The addition of the fillers changes the inner structure of the membrane, reducing the free volume in the membrane and, also, acting as barriers for the transport of sulfur dioxide from the anode to the cathode of the cell. These crossover results help to comprehend the hydrogen production results and the best performance of the composite membranes compared to the pristine PBI membrane shown in Figures 4 and 5.

Conclusions
Composite PBI membranes containing TiO2 as inorganic filler have successfully been prepared for the electrolyzer of the HyS cycle. TiO2-PBI membranes demonstrated superior performance in terms of SO2 crossover reduction, proton conductivity, electrolysis performance, and more importantly, in terms of hydrogen production. This work shows the importance of measuring and characterizing the hydrogen outlet stream, in order to determine whether a change in any of the components of the electrolyzer improves its performance. In this regard, the composite membrane 1.0 wt% TiO2-PBI membrane

Conclusions
Composite PBI membranes containing TiO 2 as inorganic filler have successfully been prepared for the electrolyzer of the HyS cycle. TiO 2 -PBI membranes demonstrated superior performance in terms of SO 2 crossover reduction, proton conductivity, electrolysis performance, and more importantly, in terms of hydrogen production. This work shows the importance of measuring and characterizing the hydrogen outlet stream, in order to determine whether a change in any of the components of the electrolyzer improves its performance. In this regard, the composite membrane 1.0 wt% TiO 2 -PBI membrane showed the best results among the studied membranes. Regarding temperature, when this parameter is increased above 120 • C, there is a change in the kinetic rates of the system with an associated reduction in hydrogen production and Faradaic efficiency. Therefore, more efforts have to be made to enhance the materials for this, trying to prevent the crossover of SO 2 to help increase temperature without losing hydrogen production efficiency.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.