Influence of the Electrode Deposition Method of Graphene-Based Catalyst Inks for ADEFC on Performance

The utilization of graphene as a catalyst support has garnered significant attention due to its potential for enhancing fuel cell performance. However, a critical challenge in electrode production still lies in the electrode preparation technologies and the restacking of graphene sheets, which can greatly impact the fuel cell performance alongside with catalyst development. This study aimed to investigate the impact of different electrode deposition methods for N-rGO-based catalyst inks on catalyst layer morphology, with a specific focus on graphene sheet orientation and its influence on the performance of alkaline direct ethanol fuel cells (ADEFCs). The dispersion behavior and ink stability of the catalysts were assessed using ultraviolet–visible light (UV-vis), ζ potential, and dynamic light scattering techniques. The morphology and physical properties of the gas diffusion electrodes (GDEs) were analyzed through Brunauer–Emmett–Teller measurements, contact angle measurements and scanning electron microscopy (SEM) combined with energy-dispersive spectroscopy. The electrochemical behavior was evaluated both ex-situ, utilizing half-cell GDE measurements, and in situ, through single-cell tests. The N-rGO-based membrane electrode assembly, comprising Pt-free catalysts and a biobased membrane, exhibited outstanding performance in ADEFCs, as evidenced by high maximum power density values and long-term durability. The N-rGO-based membrane electrode assembly has demonstrated remarkable potential for high-performance fuel cells, presenting an exciting avenue for further exploration.


INTRODUCTION
Graphene, a graphite derivative, is a two-dimensional (2D) nanosheet layer of carbon atoms that are sp 2 hybridized and arranged in a honeycomb structure.−9 There are different synthesis pathways for graphene available, whereby the Hummers method (chemical oxidation of graphene) is the most popular chemical process and results in graphene oxide (GO), which has oxygen containing functional groups and defects.These modifications change the properties of the material, from a hydrophobic surface to a hydrophilic one and from an intrinsically conductive material to an insulating one (sp 3 hybridization).−10 The modification of the carbon backbone of GO by doping with heteroatoms, such as nitrogen (e.g., NGO or N-rGO), can further increase the conductivity. 5,8,11−9 The addition of graphene to polymeric fuel cell membranes enhances the ionic conductivity, reduces the fuel permeability, increases tensile strength and improves the water retention, as already discussed in several review articles. 1,4,5,8The main focus, however, is on the development and distribution of the prepared electrodes was analyzed and the activity was determined with half-cell gas diffusion electrode (GDE) measurements.The best-performing electrodes were measured with single cell tests at different operating conditions.In accordance with the best of our knowledge, the first completely N-rGO based MEA (both catalysts and membrane) for use in ADEFCs was thus produced.In addition, the durability of the MEA was examined over time.The N-rGO-based MEA reached higher power densities and a higher long-term durability, compared to commercially available materials.

EXPERIMENTAL SECTION
In this study, N-rGO supported anodic and cathodic catalysts were synthesized and the dispersion behavior of these catalysts in various solvent mixtures and the stability of the resulting catalyst inks was evaluated.The effects of four different electrode deposition methods of the most stable and optimally dispersed catalyst ink for each catalyst on the morphology, activity and performance in an ADEFC were investigated.
The following materials and chemicals were used for the electrochemical and single-cell measurements.Ethanol (EtOH, 99.9% p.a.) and KOH (≥85%, p.a., pellets) were derived from Carl Roth (Karlsruhe, Germany).Carbon paper (Sigracet 29 BC, 0.235 mm thick), carbon cloth (ELAT -Hydrophilic Plain Cloth, 0.406 mm thick) and a commercial Pd/C catalyst (40 wt %) were delivered by Fuel Cell Store (College Station, TX, USA).Fumasep FAA-3−50 (anion-exchange membrane, nonreinforced) and a commercial Pt/C catalyst (platinum, nominally 40% on carbon black) were purchased from Fumatech (Bietigheim-Bissingen, Germany) and Alfa Aesar (Haverhill, MA, USA), respectively.A laboratory-made anion exchange N-doped graphene derivative based chitosan-membrane (CS/N-rGONRs) from a previous study was employed. 38.2.Catalyst Synthesis.The cathodic N-rGO based catalyst was synthesized as described in previously published literature.29 Therefore, the synthesis itself is not described here.The catalyst consists of 30 wt % active material (10 wt % Ag and 20 wt % Mn x O y ) on the support material N-rGO.
The anodic N-rGO catalyst also consists of 30 wt % active material (85 at.% Pd, 10 at.% Ni, and 5 at.% Bi) on the same support material (N-rGO).This 30 wt % catalyst was prepared as stated in our previous works, 25,56 with the exception of N-rGO, which was used as support material instead of Vulcan or rGO.

Preparation and Physical Investigation of the Catalyst Inks.
The preparation of stable and well-dispersed graphene catalyst suspensions is an important issue for the fabrication of electrodes. 50herefore, different solvent mixtures of water and isopropanol for both catalysts (4 mg mL −1 ) were produced with different relations The UV-vis measurements were performed with diluted inks (1:100 sample:water) in a quartz glass cuvette with a Shimadzu UV-1800 visible scanning spectrophotometer (Shimadzu, Kyoto, Japan) in a wavelength range between 200 and 800 nm at room temperature (RT) at time intervals of 0, 1, 2, 3, and 6 h.A Litesizer 500 (Anton Paar, Graz, Austria) was used for the ζ potential and DLS measurements with diluted inks (1:100 sample:water), which were directly taken after ultrasonication.The measurements were performed at RT with a scattering angle of 175°in an Omega cuvette (Anton Paar, Graz, Austria).Average values and standard deviations from three individual measurements were calculated for all measurements (UV-vis and ζ).
2.4.Electrode Fabrication.For the determination of the effects of the electrode deposition method of the N-rGO-based catalysts inks on the GDL (carbon cloth comprised the anode and carbon paper comprised the cathode), four different methods (ultrasonic spray coating, drop coating, brush coating, and roll coating) were tested, as shown in Figure 1.
The catalyst ink composition (ratio of the solvents) for all tested deposition methods was the same for each catalyst and was selected according to the UV-vis, ζ, and DLS studies.However, the catalyst content for the roll coating method was 40 mg mL −1 , since, in this production method, a paste rather than an ink was needed. 58An active area of 4 cm 2 of the GDLs, attached to a porous PTFE substrate heated to 80 °C, was coated to produce the GDEs.Care was taken to ensure that the catalyst ink was applied evenly (in multiple directions) and slowly so that the solvent could evaporate.An ultrasonic Sonotech ExactaCoat OP3 spray coater (SonoTek Corporation, USA) with a 120 kHz nozzle was employed for the spray-coating process, along with a pipet for the drop-coating process, a paint brush for the brush-coating process, and a metal roller with handle for the roll-coating process.For all cathodic electrodes, an active material loading of 0.25 mg cm −2 was achieved, whereas an active material loading of 0.5 mg cm −2 for all anodic electrodes was achieved.
2.5.Physicochemical and Electrochemical Electrode Characterization.The prepared electrodes (cathodes and anodes) were physicochemically analyzed with contact-angle measurements, the Brunauer−Emmett−Teller (BET) method and scanning electron microscopy (SEM), and their electrochemical behavior was determined with half-cell GDE experiments and single-cell tests.
SEM analysis coupled with energy-dispersive spectroscopy (EDX) was performed to determine the electrode coating and distribution, as well as the cross section with an FEI-XL20 (Philips, Amsterdam, The Netherlands) and an EDX-detector from remX GmbH (Bruchsal, Germany).Therefore, the electrodes were cut with a sharp scalpel and adhered to the sample holder with conductive carbon tape.The water contact angles (hydrophilicity) of the electrodes were measured with a goniometer (Ossila, South Yorkshire, United Kingdom) by dropping ∼3.5 μL on the surface.The analysis of the contact angles was performed with the Ossila Contact Angle software.The measurements were performed three times at RT and the average values and standard deviations calculated.The BET surface area was determined on an ASAP 2020 Micromeritics (Norcross, GA, USA) instrument by single point analysis after degassing the electrodes (1 cm 2 ) in a glass tube under vacuum.
The prepared GDEs (anodes and cathodes) were punched out in circular form (active area: 1 cm 2 ) and investigated by half-cell GDE measurements with respect to their electrochemical behavior, such as the oxidation and reduction processes of the active species (and electrochemical surface area (ECSA)) as well as the activity, with respect to ORR (cathodes) and EOR (anodes).The electrodes were labeled according to the method of manufacture and whether they were cathode or anode (with subscripts C or A).A measurement protocol from our previous work 42 with a Zahner IM6ex potentiostat coupled with a PP240 power-potentiostat (Zahner-elektrik GmbH & Co. KG, Kronach-Gundelsdorf, Germany) was applied and is therefore only briefly described here.The GDEs were measured in a Diskfix electrode holder from Bank Elektronik−Intelligent Controls GmbH (working electrode) in a three-electrode setup in 5 M KOH.A reversible hydrogen electrode (RHE) in a Luggin capillary from Hydroflex gaskatel was used as a reference electrode and a platinized titanium rod from Bank Elektronik−Intelligent Controls GmbH was used as a counter electrode.The cyclic voltammograms (CVs) were conducted after 1 h N 2 -purging and cleaning cycles with a scan rate of 10 mV s −1 in a potential range of 0.1−1.0V vs RHE with the cathodes and in a potential range of 0.05−1.50V vs RHE and 0.05−1.20 V vs RHE (ECSA determination) with the anodes.The ORR and EOR activity was evaluated with polarization curves, which were post iRcompensated (with resistance determination at each measurement point), by supplying either oxygen with a flow rate of 25 mL min −1 or a mixture of 5 M KOH and 3 M EtOH with a flow rate of 5 mL min −1 through the electrode at RT (condition I), 60 °C (condition II) or 80 °C (condition III).
The best-performing electrode, both cathode and anode in the halfcell GDE tests, was selected for the MEA, and the MEA was designated as the N-rGO based MEA.The MEA for the single cell tests was prepared by assembling the anodic GDE, the pretreated (24 h in 1 M KOH) CS/N-rGONRs AEM and the cathodic GDE together.The fuel cell tests were performed galvanostatically using the same potentiostat as described before and employed a homemade ADEFC test rig (test station) and cell, as described in detail in ref  were conducted by applying 15 mA cm −2 for 12 h at condition III and by measuring the voltage loss. 60n addition, an MEA of commercial materials: a Pd/C catalyst at the anode, Fumasep FAA-3-50 anion exchange membrane and a Pt/C catalyst at the cathode (denoted as comm.MEA) with the same active material loadings was fabricated and tested for comparison.

Physical Investigation of the Catalyst Inks.
The stability and agglomeration of the catalyst inks affect the behavior of the electrode and its fabrication. 50,52The aggregation size of the ionomer and the catalyst particles, and thus the physical and mass transfer properties of the catalyst layer, are controlled by the nature of the dispersion medium. 58Therefore, the absorbance of the catalyst particles, and thus solubility as well as stability, was determined with UV-vis spectroscopy in various solvent mixtures (isopropanol and water).Moreover, the stability or aggregation behavior was analyzed with ζ potential measurements in combination with DLS for the evaluation of the hydrodynamic radius.In Figure 2 and Table S1 and S2 in the Supporting Information, the physical investigation results for the different catalyst ink solutions can be found.
The maximum absorption (marked with the arrow in Figures 2d and 2e) of the catalyst inks was used to determine the solubility of the catalysts in the solution medium because a higher absorption intensity is the result of a higher turbidity caused by the catalyst.Moreover, Poorsargol et al. 53 used the absorption intensity to evaluate the concentration of the graphene dispersions.The peak maximum for both N-rGO based catalyst inks is located at ∼280 nm and can be attributed to the sp 2 hybridized C�C bonds. 55The content of the attached oxygen groups on the N-rGO is relatively low since, as Liu et al. 55 demonstrated, that the absorption spectra for GO demonstrates a maxima for the aromatic C−C bonds (π−π* transition) at 238 nm, which red-shifts with the reduction of oxygen groups, due to conversion from sp 3 to sp 2 coordinated carbon atoms and a small peak at ∼303 nm for the C�O bonds (n-π*), which disappears with reduction.The freshly prepared PdNiBi/N-rGO inks (Figure 2 a−black bars) indicate a constant absorption maximum for the samples 1−5, a very high value for sample 6, and subsequently, for samples 7−11, a very sharp rapid decline.The low oxygen content in the N-rGO sheets, as seen with the position of the peak maxima, causes an increase in the π−π restacking forces of the carbon atoms and thus an instability in water. 55The absorption maxima, however, is in the same area for all samples for the Ag−Mn x O y /N-rGO inks (Figure 2c black bars).Consequently, they show similar solubility behavior, regardless of the composition of the varying solvent ratio.We consider that the different active material on the N-rGO sheets, which contains transition-metal oxide, is the reason for this.Transition-metal oxides on the surface of the N-rGO sheets are able to minimize restacking, and therefore no substantial change can be seen. 61Moreover, the oxygen content or the quantity of polar functionalities has an influence on the hydrophilicity of the catalyst. 55he average hydrodynamic radius (Figure 2b and Tables S1  and S2), which describes the particle with solvation shell and was determined with DLS of all the samples, is relatively large (700−1900 nm), due to the large size of the N-rGO sheets (graphene sheets are typically in the low micromolal range, as can be seen on electron microscopy images in the literature 25,26,45 ) and the Nafion ionomer in the solution.The values are basically higher for the samples containing greater amounts of water due to the aggregation of the N-rGO sheets (van der Waals forces and π−π stacking), 55 and we assume that the hydrophilic character of Nafion and, thus, its attraction to water also has an impact.In addition, the dielectric constant of the different solvent mixtures decreases with higher quantity of isopropanol in the solution, 62 which, in turn, influences the dispersion behavior of the ionomer (distribution and dimensions of aggregates) and of the catalyst. 63he stability of dispersions can be easily described with the ζ potential, as it specifies the electrostatic repulsion between particles, whereas values of < −25 mV or >25 mV are expected to be stable. 53The stability of graphene-compounds in solution depends on the balance between electrostatic repulsion and van der Waals attraction. 54The measured ζ potential values are more negative than −25 mV for all measured samples and are therefore considered as stable.The values are negative because the Nafion ionomer or its ionic properties interact with the N-rGO sheets and transfer an effective charge onto them. 53For the PdNiBi/N-rGO samples (Figure 2a), a clear trend can be seen: the ζ potential becomes more and more negative starting from sample 1 (decreasing isopropanol content), has its lowest value at approximately samples 5 and 6 and then becomes less negative again with higher water content.The higher water content, as described before, enables the π−π restacking of the N-rGO sheets. 55The first decrease (higher potential values) may refer to the Nafion  content (hydrophilic) in the solution.For the Ag−Mn x O y /N-rGO samples (Figure 2 c), no clear upward or downward trend is observed (all samples are in a similar range), as shown previously with the UV−vis measurements, which can again be attributed to the different active material on the N-rGO sheets. 61he stability of all tested catalyst inks over time is not particularly high, as the absorption maxima in Figures 2d and  2e and Tables S1 and S2 in the Supporting Information decrease over the measured time period due to aggregation of the N-rGO sheets. 50However, sample 6 still show the highest absorption values after 6 h (gray bars in Figures 2a and 2c) and thus still the highest catalyst content in solution and the least sedimentation.The PdNiBi/N-rGO samples show a higher decrease in the absorption maxima in 6 h in comparison to the Ag−Mn x O y /N-rGO samples, due to the lower oxygen content in the catalysts and smaller metal particles and thus, the π−π stacking forces. 55,61ased on the UV−vis (high start and end absorbance values), ζ-potential (high negative value) and DLS studies (medium hydrodynamic radius, due to aggregation), a catalyst ink composition of isopropanol:water 5:5 (sample 6) was selected for the study of the different electrode deposition methods for both catalysts.
3.2.Physicochemical Electrode Characterization.The morphology of the electrode significantly influences the durability and the performance of fuel cells since the electrochemical processes occur at the catalyst layer−electrolyte interface.A high catalyst layer uniformity and distribution, together with a low aggregation quantity leads to reduced ohmic losses (and mass transport losses) and, thus, a higher power output. 64Therefore, the morphology of the catalyst layer, as well as the distribution of the active material on the N-rGO was analyzed with SEM coupled with EDX (see Figure 3, as well as Figures S1 and S2 and Table S3).In addition, the surface area was characterized with BET and the hydrophilicity with contact angle measurements (see Figure 4, as well as Table S4 in the Supporting Information).
The morphology of the prepared cathodic GDEs (Figures 3 a−d) on the carbon paper show clear differences in accordance to the electrode deposition method: the spray C has a uniform dense morphology; the drop C has large cracks but the catalyst layer is fluffier; the brush C has a uniform porous catalyst layer structure; and the roll C has a very densely packed catalyst layer.The cross-section images show that the N-rGO sheets are stacked vertically for spray C and roll C , whereas for drop C and brush C a disordered randomly arrangement can be detected, which corresponds well to observations made from the surface.
In the SEM image (SE detector) in Figure 3e of the brush C , the single N-rGO sheets show a wrinkled 2D structure and are transparent (dark); if they stick together or form agglomerates, they appear white. 50Moreover, the large size (μm-range) of the N-rGO sheets, as determined with the hydrodynamic radius in the catalysts inks with DLS, can be seen.The influence of the electrode deposition method on the morphology, as described previously, becomes more visible at higher magnification levels (SE detector images in Figure 3e and Figure S1).The destruction of graphene-based materials during ultrasonication can be an issue as described in the literature. 7,65No destruction of the N-rGO structure through the ultrasonic nozzle during the spray-coating process was observed; nevertheless, no statement can be made about the dispersion of the catalyst inks in the ultrasonic bath since all have undergone the same procedure.The active material is independent of the electrode deposition method uniformly distributed over the entire surface area (Figure 3e and Figure S1) and the atomic ratio of the elements to each other is in a similar range for all prepared cathodic GDEs (see Table S3 in the Supporting Information).
The influence of the deposition method on the morphology of the electrode is even more evident in the SEM images of the anodes (Figures 3f−i), due to the use of a different GDL (carbon cloth vs carbon paper): the strands of the carbon cloth are evenly coated with catalyst only in spray A ; brush A shows coated and uncoated areas, as well as agglomerates; the agglomeration and noncoating is slightly more pronounced in drop A ; and roll A shows very poor coating on the surface and agglomerates under the surface.In the case of drop A , brush A , and roll A , large quantities of catalyst are pressed or flow through the carbon cloth mesh, as can be seen in the crosssection images.Conversely, this does not happen with spray A , where a fine catalyst layer can be seen on the surface of the threads.Jhong et al. 64 also observed, when comparing different electrode deposition methods on carbon cloth for Pt/C catalyst cathodes for fuel cells, that agglomerates are formed when using a brush for deposition, this is due to agglomerate formation and growth inside the brush, which is then deposited.In the SE detector images in Figure 3j and Figure S2 in the Supporting Information of the anodic GDEs, a clear difference to the cathodes becomes apparent, i.e., the N-rGO sheets are much more restacked (most clearly at drop A and roll A ).The different active material, pure metal (anodic) in comparison to the transition-metal oxide hybrids (cathodic), results, as previously observed with the UV-vis and ζ potential measurements, in an increase in the π−π restacking forces of the N-rGO sheets. 61The EDX analyses resulted in a uniform distribution of active material and in comparable atomic ratios for all prepared anodic GDEs (see Figure 3j, Figure S2, and Table S3).
In addition to the visual investigation of the surface morphology, BET analyses (Figures 4a and 4b and Table S4) were used to describe the specific surface area (S BET ) of the GDEs and the blank GDLs for comparison.The following trend for the S BET values, which is in accordance with the observations from SEM, for the cathodic GDEs can be observed: spray C (42.2 m 2 g −1 ) > drop C (34.0 m 2 g −1 ) ≈ brush C (34.7 m 2 g −1 ) > roll C (25.6 m 2 g −1 ), whereas, the S BET values are smaller for the anodic GDEs: spray A (14.3 m 2 g −1 ), drop A (25.9 m 2 g −1 ), brush A (12.6 m 2 g −1 ) and roll A (31.0 m 2 g −1 ).This can be attributed to the lower S BET of the blank A in comparison with blank C and the higher restacking degree.For spray A and brush A , no increase is observed after application, which is probably due to the coating of the pre-existing threads.Drop A and roll A show an enlargement, which is assumed to be the result of the large agglomerates between the threads, which were observed with SEM analysis.The BET analysis shows that, in every case (no matter whether cathode or anode, or which electrode deposition method was used), restacking has occurred, because, compared to the pure support material, a reduction was observed.Nosan et al. 57 measured an S BET value of 74 m 2 g −1 for the same N-rGO material (not on a GDL).
The hydrophilicity requirements on the cathodic and anodic electrodes in the ADEFC are disparate.The catalyst layer at the cathode should be more hydrophobic, because of the gaseous oxygen, whereas the anode should be more hydro-philic because of the liquid ethanol and KOH mixture. 42,66The hydrophilicity of the GDE surfaces was thus investigated with contact angle measurements (see Figures 4c and 4d, as well as Table S4).The prepared cathodes become more hydrophilic in comparison to the blank GDL, whereas, when it came to the anodes, the hydrophilicity is reduced after deposition of the catalyst layer.Similar to the observations made previously in the catalyst ink measurements, this is caused by the different active material, more specifically on the oxygen content, on the N-rGO sheets. 55In addition, it can be observed that, regardless of active material (cathodic or anodic), the electrode deposition method and thus the emerging electrode morphology influences the hydrophilicity: the drop coating and roll coating production methods produce more hydrophilic electrode surfaces than ultrasonic spray coating or brush coating.The surface hydrophilicity or hydrophobicity is influenced by morphology (and also surface energy), which, in turn, depends on anchoring and surface tension. 67.3.Electrochemical Electrode Characterization.The prepared electrodes were electrochemically characterized with the use of half-cell GDE measurements, to determine the oxidation and reduction processes (with cyclic voltammetry), as well as the ORR and EOR activity (with polarization curves).Half-cell GDE measurements are a helpful tool to investigate the electrodes at high current densities with a viable catalyst layer structure and three-phase boundary.42,68 Therefore, the influence of the electrode deposition method on the electrochemical behavior can be demonstrated.In addition to the half-cell GDE measurements, the best-performing electrodes were used for single-cell tests to determine their performance and durability.
In the cyclic voltammogramms (CVs) of the cathodic electrodes in Figure 5a, the oxidation and reduction peaks for manganese oxide (primarily visible: formation of MnOOH at 0.76 V vs RHE and 0.58 V vs RHE), but not for Ag, since they appear at a higher potential, can be seen. 29,69All of the produced cathodes show the same peaks; however, roll C exhibits hysteresis in the baseline (high slope), due to charging currents and a higher resistance, 70 which is indicated by the observed dense electrode and the restacking of the N-rGO sheets with SEM.The estimated double-layer capacitance (C dl ) between 0.2−0.3V vs RHE, where no faradaic processes occur, allows a statement about the surface area of the catalyst, since it is proportional to all conductive components. 42,71The trend of the C dl values for the four different electrodes are consistent with the trend of the physical surface area measured with BET: spray C (94 F g −1 ), drop C (72 F g −1 ), brush C (78 F g −1 ), and roll C (59 F g −1 ).Thus, the influence of the electrode deposition method on the morphology (distribution and agglomeration) on the catalyst material could be additionally supported electrochemically.
The iR-compensated ORR polarization curves for the different conditions, shown in both Figure 5b−d and the values obtained therefrom in Figure 5e and Table S5, show a clear performance trend of the four prepared cathodic GDEs.The performance of all electrodes increases between condition I (25 °C) and condition II (60 °C) and stagnates between condition II and III (80 °C).The brush C shows the highest current density (j) values at 0.7 V vs RHE in comparison to the other three electrodes.The onset potential (E onset ) is in opposition for brush C , drop C , and roll C , and it is independent of the condition for all three in a similar range.Spray C , in contrast, shows significantly inferior values.The performance of the roll C electrode improves more with condition (higher temperature) in comparison to that of brush C or drop C , since reactant diffusion is increased with temperature, and the limitation caused by restacked N-rGO sheets (observed with SEM) is eliminated to some extent.The same can also be  S5 in the Supporting Information.The Tafel slope describes the kinetics of the observed reaction (how facile), whereas the product of n (number of electrons) and α (transfer coefficient) describes the efficiency of the catalytic interface. 72The Tafel slopes are the lowest and the nα values are the highest for the brush C electrode, whereas the opposite applies to spray C .Therefore, the optimal catalyst layer utilization for the cathodic GDEs was achieved with the brush C electrode.The low performance of the spray C electrode is a result of the dense electrode layer structure.The observed restacking of N-rGO sheets with SEM and S BET leads to a reduction of accessible catalytic active sites, surface area, and the diffusion pathways and thus leads to an efficiency loss. 9,45Grigoriev et al. 47 observed that the graphene sheets irreversibly agglomerated and horizontally stacked during electrode production with a Pt/rGO catalyst for the polymer electrolyte membrane fuel cell (PEMFC), which resulted in lower activity.
The CVs of the anodic electrodes displayed in Figure 6a show the oxidation of Bi to Bi 2 O 3 in the anodic scan at 0.9 V vs RHE and the reduction of PdO to Pd in the cathodic scan between 0.9−0.4V vs RHE. 12,13,17,25The oxidation and reduction peaks of Ni (Ni(OH) 2 ↔ NiOOH) are not visible in this potential range, as they occur between 1.2−1.5 V vs RHE; 25 however, they can be found in Figure S3 in the Supporting Information.The PdO reduction peak was used to calculate the electrochemical active surface area (ECSA), as described in the literature. 13,15,20The electrodes were manufactured with the same metal loading; thus, any change in the ECSA can be attributed to the accessibility of the catalytic active material.Therefore, with spray A , the best active catalyst material utilization can be achieved, as it clearly shows the greatest ECSA (426 cm 2 mg −1 ) compared to the other three electrodes or electrode deposition methods (drop A : 109 cm 2 mg −1 , brush A : 54 cm 2 mg −1 and roll A : 49 cm 2 mg −1 ).The high ECSA is due to the good accessibility of the active sites, as well as the low agglomeration of the catalyst material, 64 as observed by the SEM images.The roll A CV shows similarly as roll C a modification of the flat baseline, and thus this phenomenon can be attributed to the electrode deposition method.The estimated C dl at 0.44 V vs RHE follows the same trend as the ECSA: spray A (135 F g −1 ), drop A (36 F g −1 ), brush A (20 F g −1 ) and roll A (28 F g −1 ).The C dl for the anodes, in contrast to the cathodes is not consistent with the S BET values, which can be attributed to the different surface structure of the substrate used and the observed restacking and agglomeration.This results in nonutilizable active material (enclosed between sheets) and thus differences in the various values, that describe the different surface areas (as described before) of ECSA, C dl , and S BET .
The activity of the prepared anodic GDEs increases for nearly all the electrode deposition methods with increasing temperature (condition I to III) due to improved EOR kinetics 42,66 and results in higher j, lower E onset , lower OCP, higher nα values, and lower Tafel slopes, as shown in Figure 6b−e and Table S5 in the Supporting Information.In each of the tested EOR conditions, spray A definitely shows the best performance and activity and thus the highest catalyst layer utilization.The lower performance of drop A , brush A , and roll A , when compared to spray A , depends on the lower ECSA values of the electrodes, caused by the observed agglomeration, restacking and the inaccessible active catalyst material, 64 as described before.The performance of the roll A electrode improves significantly (as previously observed with roll C ) with higher temperature, due to increased reactant diffusion.The performance of drop A and brush A is quite comparable for all tested conditions, due to the similar morphology observed with SEM.
Remarkably, the influence of the electrode deposition method has quite the opposite effect in terms of activity for the cathode and anode, e.g., the anode prepared with ultrasonic spray coating achieved the highest performance.However, the cathode following the same method was clearly the worst performing.This is due to the use of different GDLs with unequal substrate surface structures and the different aggregate state of the reactant and thus the hydrophilic requirements, as previously mentioned.
The electrodes for the MEA used for the single cell tests were thus brush C and spray A .The maximum power density (p max ) values for the different conditions can be found in Table 1.
The p max values increase from condition I to III, due to increasing temperature and therefore, the membrane conductivity, the electrode kinetics, and the mass transfer properties are enhanced. 42,66The N-rGO MEA is superior to the comm.MEA under any operating condition.The higher performance of the N-rGO based MEA, in comparison to the comm.MEA (Figure 7a−c), can be attributed to several factors: (i) the inclusion of N-rGONRs in the CS-membrane due to the resulting hydrophilic regions (oxygen groups), which enable easier hydroxide transport and reduced ethanol crossover due to hydrophobic domains (sp 2 carbon atoms), 37,38 (ii) the ethanol tolerance of the cathode catalyst due to the utilization of Ag−Mn x O y in contrast to Pt, 29 (iii) the adatoms (Ni and Bi) in the Pd-based anodic catalyst, due to favored OH − adsorption on the surface, 25,56 and (iv) the utilization of N-rGO as catalyst support, due to better distribution of the active material and the improved oxidative removal of EOR intermediates. 8,22Moreover, the improvement in ethanol crossover suppression and the ethanol tolerance of the cathode is also reflected in the much higher OCV of the N-rGO MEA compared to the comm.MEA (e.g., condition I: 0.909 V vs 0.705 V, respectively).The OCV is lower for the comm.MEA, as the Pt/C catalyst shows EOR activity, and thus in the case of ethanol crossover, leads to mixed potentials. 29,39,42he durability study of the N-rGO based MEA (Figure 7e) resulted in a smaller (half as much) degradation rate (d) of 0.3 mW cm −2 h −1 for the 12 h measurement period in comparison to the comm.MEA (Figure 7d), which was 0.6 mW cm −2 h −1 .A higher ethanol crossover rate through the membrane leads to unused fuel for oxidation and thus activity losses, which is reduced through the incorporation of N-rGONRs in the membrane, as described previously. 37,38The ethanol crossover from the anode to the cathode results moreover, in a larger potential loss for Pt-catalysts, in contrast to the Ag−Mn x O y /N-rGO catalyst. 29Therefore, we expect that the visible voltage fluctuations, which are clearly increasing for the comm.MEA are due to the occurrence of an ethanol crossover and to the accumulation of ethanol oxidation intermediates on the anodic catalyst surface.This is because, in contrast to the N-rGO MEA, there is no N-rGO, Ni, and Bi present, which supports the oxidation removal of intermediates. 8,21,22,25,56In addition, the morphology of the electrodes was analyzed after the durability study with SEM and EDX, as shown in the Supporting Information (Table S6, Figure S4 and S5).Compared to the freshly manufactured electrodes, there is nearly no difference in the morphology and the atomic ratio of the elements to each other, as well as the even distribution of the active material is again observed for both GDEs.Hence, the degradation in the last 6 h of the measurement is remarkably reduced for the N-rGO based MEA and leads to a small d of 0.03 mW cm −2 h −1 , in contrast to the comm.MEA.Hou et al. 60 observed a d of 0.027 mW cm −2 h −1 during a  durability study of 336 h.Thus, a very high long-term durability of the N-rGO based MEA can be assumed.The comm.MEA was prepared as a reference for comparison, as the literature data of graphene compounds in the ADEFC is rare and complex: the conditions are not reported as sufficient or are different from the conditions used in this work.However, a literature performance comparison of various graphene compounds containing MEA components can be found in Table 2.The competitiveness of the produced N-rGO based MEA is highlighted considering the use of completely Pt-free catalysts with low loadings and the biobased membrane material with a high p max of 62.6 mW cm −2 at 80 °C.

CONCLUSION
The influence of the electrode deposition method (ultrasonic spray coating, drop coating, brush coating, and roll coating) of N-rGO based catalyst inks�Ag−Mn x O y /N-rGO for the cathode and PdNiBi/N-rGO for the anode�on the morphology and thus on the performance in the ADEFC was successfully determined, whereby restacking of graphene sheets has a major effect.Therefore, the dispersion behavior and the stability of the catalyst inks was evaluated first in various solvent mixtures.It was shown that the nanoparticles on the N-rGO sheets affects the π−π restacking forces and thus the stability of the ink.The most optimal ratio of isopropanol to water for a stable and well dispersed ink is 1:1.The properties of the two different catalysts (anode or cathode), such as a tendency toward restacking or agglomeration, were also observed in the produced GDEs, which influenced the electrochemical activity: the brush coated cathode and the ultrasonic spray coated anode achieved the highest levels of performance.The different GDLs and the different requirements for hydrophobicity and hydrophilicity for the anode and cathode affect both the structure and morphology of the catalyst layer and which method of electrode deposition is most suitable.The completely N-rGO based MEA, which is Pt-free and includes a biobased membrane, produced with the best-performing cathode and anode reached maximum power densities of 62.6 mW cm −2 at 80 °C�almost triple the power output in comparison with the commercial AEM and catalysts.Moreover, the incorporation of graphene-compounds into the MEA resulted in good longterm durability, with a small degradation rate of 0.03 mW cm −2 h −1 compared to the commercial MEA (double).Therefore, the influence of the active material on the N-rGO sheets on the restacking and the impact of the surface texture of the GDL during electrode deposition were demonstrated, paving the way for further improvement in the use of graphene-based materials for fuel cell electrode production.
59.A mixture of 3 M EtOH and 5 M KOH at a flow rate of 5 mL min −1 was used as fuel at the anode while oxygen at a flow rate of 25 mL min −1 was used at the cathode.Three different operating conditions were tested: condition I (RT, pure O 2 ), condition II (60 °C, humidified O 2 ), and condition III (80 °C, humidified O 2 ).Furthermore, constant-current discharging tests (durability tests)

Figure 1 .
Figure 1.Schematic illustration of the four different electrode deposition methods: ultrasonic spray coating, drop coating, brush coating, and roll coating.

Figure 3 .
Figure 3. SEM images of the electrode coating (cross-section in the insets−scalebar = 100 μm): (a) spray C , (b) drop C , (c) brush C , (d) roll C , (f) spray A , (g) drop A , (h) brush A , and (i) roll A ; and EDX-mapping of (e) brush C and (j) spray A .

Figure 4 .
Figure 4. BET surface area results of (a) the cathodes and (b) the anodes; and water contact angles (θ) of the electrodes (c) cathodes, and (d) anodes.

Figure 5 .
Figure 5. Half-cell GDE measurements of the produced cathodes (a) CVs; iR-compensated ORR polarization curves (Tafel plots in the insets: η = overpotential, j = current density) at (b) condition I, (c) condition II, and (d) condition III; and (e) comparison of the onset potentials (E onset ) at −10 mA cm −2 and current density (j) at 0.7 V vs RHE for the different conditions.

Figure 6 .
Figure 6.Half-cell GDE measurements of the produced anodes (a) CVs (ECSA in the inset); iR-compensated EOR polarization curves (Tafel plots in the insets: η = overpotential, j = current density) at (b) condition I, (c) condition II, and (d) condition III; and (e) comparison of the onset potentials (E onset ) at 10 mA cm −2 and current density (j) at 0.65 V vs RHE for the different conditions.

Figure 7 .
Figure 7. Polarization (unfilled symbols) and power density curves (filled symbols) of the ADEFC tests at (a) condition I, (b) condition II, and (c) condition III.The discharging curves of the durability study for (d) the comm.MEA and (e) the N-rGO-based MEA.
(a) spray C , (b) drop C , and (c) roll C ) (Figure S1); EDX mapping of the produced anodes ((a) drop A , b) brush A and c) roll A (a) SEM images of the electrode coating and (b) EDX mapping of the cathodic electrode of the N-rGO based MEA after the durability study (Figure S4); (a) SEM images of the electrode coating and (b) EDX mapping of the anodic electrode of the N-rGO based MEA after the durability study (Figure S5)PDF)

Table 1 .
Maximum Power Density Results of the Single Cell Tests for Conditions I−III