Advanced catalyst layers for PEM fuel cell produced by inkjet printing: Catalyst layer optimisation and comparison with ultrasonic-spray-coated one

Advanced catalyst layers for PEM fuel cell produced by inkjet printing: catalyst layer optimisation and comparison of performance with ultrasonic-spray-coated one – Supplementary Information T. Zubkovaa,*, K. Heinricha, M. Halab, M. Prokopb, T. Jedlickab, K. Bouzekb, A. Willerta, R. Zichnera aFraunhofer Institute for Electronic Nano Systems, Technologie-Campus 3, 09126 Chemnitz, Germany bDepartment of Inorganic Technology, University of Chemistry and Technology Prague, Technicka 5, Prague 6, 166 28, Czech Republic * Corresponding author: Tatiana Zubkova, email: [email protected], telephone: +49 371 45001-403 Keywords: PEM, fuel cell, catalyst layer, crack mitigation, inkjet printing, ultrasonic spraying 1 EXPERIMENTAL 1.1 Inkjet printing Fig. S1. Inkjet deposition process (example for 400 dpi). Black – first pass, red – second pass, green – third path, grey – last path, together comprising one printed layer. The resolution of the printing can be defined for the x (parallel to printing direction) and y (perpendicular to printing direction). The angle of the printhead in respect to the printing direction was set to 90°. Correspondingly, perpendicular to the print direction, the distance between drops in one path of the printhead is defined by the native resolution of the printhead and comprises 254 µm. To reach a desirable resolution y, the printhead had to take several paths. In parallel to the print direction x, the resolution of the printing relates to the frequency of the jetting and the speed of the printhead movement. The printhead was set to bi-direction jetting mode, where jetting occurred in forward and back movement of the printhead. Video 1: Deposition of the patterned catalyst layer by IJP. 1.1.1 Inkjet-printed CCMs Before the deposition, the protective film of the membrane was detached, while the support film remained attached. The membranes were placed directly onto the printing table with activated vacuum. Multiple overprints were done to reach 0.1 mgPt cm−2, 0.2 mgPt cm−2, and 0.3 mgPt cm−2 on the cathode side. The final drop space (DS) of 36.3 µm in each layer corresponded to 700 dpi (x = y). The printing direction was rotated by 90° after every two layers. The printing table with activated vacuum function was heated to 50 °C. The printed cathode was dried in an oven for 20 min at 80 °C. After this, the support film of the membrane was detached and printing was realized on the other side with the same printing conditions, so the total Pt loading of each CCM comprised 0.2 mgPt cm−2, 0.4 mgPt cm−2, 0.6 mgPt cm−2. The distance between the printhead and membranes was kept at 4 mm for the cathode and 7 mm for the anode side. An increased in the distance to the anode side was found necessary to avoid collision between the printhead and the substrate in the case of membrane swelling. The cathode layers were printed one after each other without additional delays. During anode printing, the membrane was dried on the printing table shortly after each printed layer to avoid excessive swelling. The obtained CCMs were placed in oven at 80 °C for 30 min for final drying. Table S1. Summary of IJP printing parameters for CCM cathodes and ink A. Loading, mgPt cm-2 In-scan resoluti on x, dpi Crossscan resoluti on y, dpi Algorith m of the drop depositi on** DS (in print / layer) in x, µm Number of layers Speed of printhead, mm s−1 Distance to the printhead , mm T of the printing table, °C IJP_0.1 700 700 2 72.6/36.3 7 50.8 4 50 IJP_0.2 700 700 2 72.6/36.3 14 50.8 4 50 IJP_0.3 700 700 2 72.6/36.3 21 50.8 4 50 * * - number 2 indicates, that only every second pixel was printed in each path of the printhead. To maintain the necessary density of pixels, a second nozzle must print additional path. 1.2 Ultrasonic spray coating Fig. S2. Pattern for ultrasonic deposition. Black – first pass, red – second pass, together comprising one printed layer. Video 2: Deposition of the catalyst layer by USC. 1.3 Electrochemical test 1.3.1 Fuel cell and test station assembly Prior to fuel cell testing, CCMs were activated in demineralized water at 80 °C for 2 h. Membrane electrode assembly (MEA) was completed directly in the testing cell without hotpressing. Sigracet 22BB carbon paper of 215 μm thickness, containing polytetrafluorethylene (PTFE)/C microporous layer (MPL) on one side, were cut into 5 cm × 5 cm squares and used as GDLs. The test cell was a pneumatic AirCell LT (LeanCat) with a 5 cm × 5 cm active area. The gas distribution plates were made of 14 mm thick high-density graphite, featuring a dual serpentine flow-field design with 1 mm wide channels and ribs, and a depth of 0.5 mm. Before assembly, the graphite plates were cleaned with demineralized water and ethanol. A 1 mm thick expanded PTFE sealing was placed in the groove along the active area perimeter. New seals were used for each sample to avoid deformation. The activated CCM was then placed onto the GDL, covering the active area. The anode gas distribution plate, including the other GDL adhered to the plate by water surface tension, was positioned on top, aligned using PTFE pins in the cathode plate. The top part of the cell with gold-plated Cu current collectors and heating cartridges was placed over the assembly. It included a pneumatic mechanism that ensures uniform compressing pressure of 5 bar (equal to force of 1.25 kN) across the cell area. Hydrogen (purity 99.95 vol.%, SIAD) was supplied to the anode, and oxygen (purity 99.5 vol.%, SIAD) to the cathode. The cell operated at 80 °C, with inlet gases preheated to 90 °C and maintained at 100 % relative humidity using Nafion tubing humidifiers (Permapure). An overpressure of 0.5 bar was applied using a KBP1F0A4A5A20000 (Swagelok) regulator. Gas flow rates were controlled by two EL-FLOW Select mass flow controllers (Bronkhorst, model F201CV5K0RAD22V). A Rigol DL3021A 150V/40A/200W DC electronic load was used for initial cell break-in. The assembled cell and testing station are shown in Fig. S3. For load characteristics, electrochemical impedance spectroscopy (EIS), and I-U curves, a Solartron ModuLab XM ECS Electrochemical Test System with 100 A booster (Ametek) was used. Fig. S3. (Left) Single cell test station. (Right) Detail of the pneumatic cell. 1.3.2 Testing protocol A break-in process was used to achieve optimal fuel cell performance. After cell assembly, the current load was increased to 1 A cm−2, ensuring that the voltage remained well above 0.4 V. After reaching steady state at a 1 A cm−2, the cell was conditioned at this current density for additional 18 h. The cells were operated at 80 °C at all times. Gas flow rates were adjusted based on the current drawn from the cell, calculated using Faraday's law, the ideal gas equation, and reaction stoichiometry, with excess coefficients  of 1.5 for H2 and 3 for O2. The load curves were recorded following these steps: • Galvanostatic conditioning at 1 A∙cm−2 for 120 s, • Open Circuit Voltage (OCV) scan for 30 s, • Linear-sweep voltammetry from 0 to 3 A cm−2 (or for U > 0.2 V) at 0.2 A s−1 with gas flows set for 3 A cm−2. EIS was used to measure the ohmic and polarisation resistances of the samples. The procedure included the following: • Galvanostatic conditioning at 1 A cm−2 for 120 s, • OCV scan for 30 s, • EIS measurement at current densities of 0.2, 0.5, 0.7, 1.0, 1.2, and 1.5 A cm−2, with 120 s of pre-polarisation, followed by 60 s of EIS at frequencies from 10,000 to 0.1 Hz with a 5 % amplitude of the current value. The EIS data were deconvoluted by regressing a model for impedance consisting of an Ohmic resistance in series with Voigt elements implemented in Measurement Model 1.8 [1]. The analysis has shown three (in selected cases four) time constants. The first time constant located at ~2·10−5 s is related to hydrogen oxidation reaction. The second time constant with a value of ~3·10−4 s is related to ion transport in CLs. The third constant at ~1·10−3 s then corresponds to oxygen reduction reaction. The potential fourth constant at ~5·10−2 s occurring in selected cases is caused by diffusion of gases to the reaction site [2,3]. The equivalent circuit used to fit the EIS data thus consists of three (four) Voigt elements and can be found in Fig. S4. Several data points at very low (up to 1 Hz) and very high frequencies (above 8 kHz) were disregarded in the analysis to reduce the noise. Then the Measurement Model software tool was used with the following parameter setup: • Fit type – Complex • Number of Simulations – 1,000 • Weighting - Modulus with value α equal to 1 The number of time constants was set to 3 or 4, ensuring that the 95 % Confidence Interval of time constants and resistances never increased above 100 % (meaning the standard deviation was smaller than the fitted value). Finally, the resistances were grouped according to their time constants. All resistance values with errors can be found in Fig. S5. Fig. S4. Equivalent circuit for three time constants, RO – Ohmic resistance, Ri – polarisation resistance, Ci – capacitor, τi – time constant, i = {1, 2, 3}. The U-I curve measurement procedure for the assessment of electrochemical surface area of cathode (ECSA) included: • Cathode purification with nitrogen (100 cm³ min⁻1) and anode flushing with hydrogen (100 cm³ min⁻1), measuring OCV for 15 min, until reaching steady state at ~0.1 V, • Baseline current measurement at 0.4 V for 120 s, • Potential cycling between 0.05 V and 1.2 V at 20 mV s⁻1 for three cycles. The electrochemical surface area was calculated using in-house developed program in MATLAB (MathWorks) by subtracting the baseline current, applying a linear fit in the 0.35- 0.6 V range to compensate for the hydrogen crossover and charging current, and performing numerical integration from time at 0.1 V to time at 0.4 V with the "trapz" function. This value was divided by the data collection rate r, electrode surface area 𝐴, charge of adsorbed hydrogen monolayer 𝑞H,ML, and absolute Pt loading 𝑚C,Pt, as shown in Eq. (S1). (S1) 2 RESULTS AND DISCUSSION Fig. S5. Polarisation resistance analysis with standard deviation error bars. A) resistance associated to the first time constant located at ~2·10−5 s related to hydrogen oxidation reaction. B) resistance related to the second time constant with a value of ~3·10−4 s related to ion transport in CLs. C) resistance related to the third time constant at ~1·10−3 s corresponding to oxygen reduction reaction. D) fourth time constant at ~5·10−2 s occurring in selected cases caused by diffusion of gases to the reaction site. Fig. S6. CLs on membranes printed with ink B at varied drop spaces and number of layers: A) 1,000 dpi, DS = 25.4 µm, 4 layers, B) 1,000 dpi with implemented algorithm of drops order, 4 layers, D) 700 dpi with algorithm of drops order, 8 layers. Loading of the samples 0.13 ± 0.02 mgPt cm−2. Schematic: white circle – first order of drops, blue circle – second order of drops. Fig. S7. Inkjet-printed catalyst layers on membrane: A) printing direction changed to 90° after each layer, B) support film from the back side of the membrane is detached before printing. Print resolution 1,000 dpi with algorithm of drop order, total 4 layers, loading 0.13 mgPt cm−2. Pt maps 20 mm x 20 mm.