Exploring the process design and 
parameters for a fuel cell membrane 
electrode assembly by decal transfer 
 
Master’s thesis in Innovative and Sustainable Chemical Engineering 
 

 
 
Dylan Schulz 
 
 
 
DEPARTMENT OF PHYSCIS 
CHALMERS UNIVERSITY OF TECHNOLOGY 

Gothenburg, Sweden 2023 

www.chalmers.se 



Abstract 
 

Proton exchange membrane fuel cells (PEM FCs) are a promising key component in a renewable 

energy infrastructure. At their heart is the membrane electrode assembly (MEA), where both the 

hydrogen oxidation and oxygen reduction reactions occur. Decal transfer is an effective method for 

fabricating MEAs, as it is highly scalable and noted for its low interfacial resistance. In this thesis, the 

main processing parameters for the decal transfer – temperature, pressure and transfer time – are 

varied to explore the parameter space available for MEA production. Based on these results, in-situ 

performance tests using polarization curves and high frequency resistance (HFR) measurements are 

performed in a 2k experiment matrix to study the effects of these parameters on the MEA 

performance. Further experiments are conducted for different proton exchange membranes and for 

in-house produced cathodes to understand the effects of these materials on the performance and 

processing parameter space. Further studies are performed to explore the glass transition 

temperatures of the membranes, to understand the real temperature profiles of the MEA during the 

transfer, and to explore the morphological changes undergone by the electrode during transfer. The 

results of this thesis most notably illustrate the sets of parameters that can result in a good transfer, 

demonstrate the effects of these parameters on the transfer quality and MEA performance. 

  



Acknowledgements 
 

No project of significance is undertaken alone, and this thesis is no exception. I would like to take this 

opportunity to thank those who have helped me during my thesis. 

At PowerCell, Parinaz Mikaeili has been a wonderful supervisor, helping me with everything from using 

equipment to discussing experiment designs and results. Gabor Toth has been instrumental in 

understanding results and helped me to learn about theory and how to evaluate measurements. Lukas 

Hansen and Kalle Genheden enabled me to conduct the performance measurements integral to this 

thesis. Felix Ernst and Marika Männikkö helped me with the onboarding and getting to know the lab, 

equipment and transfer process in the first stages of my thesis. Emma Ulberstad helped me with the 

preparation of in-house electrodes for my experiments. 

At Chalmers, Björn Wickman, my examiner, facilitated all contact with the university and helped 

connect me to colleagues with a DSC machine. These colleagues are Jan Svensson and Mikael Käll, 

without whom I could not have obtained glass transition temperature measurements for my 

membranes. 

A huge thanks goes to all these wonderful people, without whom I could not have achieved nearly as 

much in my thesis. 

  



List of Figures 
Figure 1: Schematic diagram of a PEM FC. ............................................................................................. 3 

Figure 2: Schematic overview of 3-, 5- and 7-layer MEAs. ..................................................................... 4 

Figure 3: Example of a DSC graph for measuring the glass transition temperature of a membrane. .... 7 

Figure 4: General circuit diagram for in-situ subscale testing. ............................................................... 7 

Figure 5: Example of conditioning data for a well-functioning MEA, from MEA D10 manufactured 

under T=200C and p=300 N/cm2 for t=300 s ......................................................................................... 8 

Figure 6: Example of conditioning data for a poorly functioning MEA, from MEA D12 manufactured 

under T=160C and p=300 N/cm2 for t=300 s. ........................................................................................ 9 

Figure 7: Example of a polarization curve with the OCV and the kinetic, ohmic and mass transport 

regions marked. ...................................................................................................................................... 9 

Figure 8: A Nyquist plot, obtained by measuring the complex EIS resistance between 5000 and 100 

Hz. ......................................................................................................................................................... 11 

Figure 9: Schematic overview of the materials in the MEA, as well as supplementary materials for the 

decal transfer. ....................................................................................................................................... 12 

Figure 10: Structure of an Ionomer (Nafion). ....................................................................................... 13 

Figure 11: An In-house prepared cathode, visual inspection using a light board. ............................... 14 

Figure 12: Supplier A electrode decals with visible defects. ................................................................. 15 

Figure 13: DSC results for membranes A, B, C and D. ........................................................................... 18 

Figure 14: Assembly plates. .................................................................................................................. 21 

Figure 15: Assembly plate damaged instead of the hot press. ............................................................. 21 

Figure 16: Pressure distribution measurement general setup. ............................................................ 22 

Figure 17: Initial prescale pressing test including cellulose pressure distribution layers, with four 2x2 

MEA samples, showing areas with and without electrodes. ................................................................ 22 

Figure 18: pressure distribution in the hot press for the four-cell array. ............................................. 23 

Figure 19: Pressure distribution in the hot press for a 33 cm^2 sample. ............................................. 23 

Figure 20: Internal temperature measurement during a decal transfer. ............................................. 24 

Figure 21: Temperature course of the MEA and hot plates during a decal transfer. ........................... 25 

Figure 22: Pressure distribution experiment setup for multiple MEAs. ............................................... 26 

Figure 23: The pressure distribution during transfer in a stack of multiple MEAs. .............................. 26 

Figure 24: Localized electrode defects appear on the pressure film. ................................................... 27 

Figure 25: Internal temperature course during transfer for one MEA and 8 MEAs. ............................ 27 

Figure 26: Reflexion behavior for (a) the raw electrode, (b) the p0 annealed electrode and (c) the 

p300 annealed electrode. ..................................................................................................................... 30 

Figure 27: MEA transfer result using p300 annealed electrodes.......................................................... 31 

Figure 28: Experiment design matrix for exploring the space of available processing parameters. .... 33 

Figure 29: Dark Green: A successful transfer with no residual. ............................................................ 34 

Figure 30: Light Green: A successful transfer with some residual. ....................................................... 34 

Figure 31: Yellow: A transfer with high residual. .................................................................................. 34 

Figure 32: Red: A failed transfer. .......................................................................................................... 34 

Figure 33: Process parameter space for M2: Supplier A anode and cathode with membrane A. ....... 36 

Figure 34: Process parameter space for M3: Supplier A anode and cathode with membrane B. ....... 37 

Figure 35: Process parameter space for M4: Supplier A anode and cathode with membrane C. ....... 37 

Figure 36: Process parameter space for M5: Supplier A anode and cathode with membrane D. ....... 38 

Figure 37: Process parameter space for M6: Supplier A anode and membrane A with in-house 

cathode 1. ............................................................................................................................................. 39 



Figure 38: Experimental setup of the MEAs in the series stacking experiment for in-situ performance 

testing. .................................................................................................................................................. 43 

Figure 39: Measurement points for the in-situ performance investigation of process parameters. ... 43 

Figure 40: Polarization curves for the five statistical repeats at operating point M2D4. ..................... 44 

Figure 41: EIS and HFR resistance values for the statistical repeats. ................................................... 45 

Figure 42: Reduced polarization curve for the statistical evaluation. .................................................. 46 

Figure 43: Reduced polarization curve for the statistical evaluation, normalized to the kinetic point.

 .............................................................................................................................................................. 47 

Figure 44: Conditioning curves for the series stack in-situ experiments. ............................................. 48 

Figure 45: HFR for the series stack experiments. ................................................................................. 48 

Figure 46: Polarization curve for the series stack tests. ....................................................................... 49 

Figure 47: Normalized polarization curve for the series stack tests. .................................................... 50 

Figure 48: Conditioning curves of the MEAs used for the 2k factorial investigation of process 

parameters. ........................................................................................................................................... 51 

Figure 49: EIS resistance data for evaluating processing parameters. ................................................. 52 

Figure 50: HFR data for evaluating processing parameters. ................................................................. 52 

Figure 51: Polarization curves for the variation of transfer temperature. ........................................... 53 

Figure 52: Polarization curves for the variation of transfer pressure. .................................................. 54 

Figure 53: Polarization curves for the variation of transfer time. ........................................................ 55 

Figure 54: polarization curves of the best performing MEAs in the parameter spacing test. .............. 57 

Figure 55: Conditioning curves for the material variation experiment. ............................................... 58 

Figure 56: HFR and EIS resistance for the material variation. .............................................................. 59 

Figure 57: Polarization curves for the material variation experiment. ................................................. 59 

Figure 58: Normalized polarization curves for the material variation experiment. ............................. 60 

Figure 59: Conditioning curves for the in-house electrode MEAs with 10, 50 and 200 kWs dispersion 

energy input. ......................................................................................................................................... 62 

Figure 60: Conditioning curves for the in-house electrode MEAs with the noise removed................. 62 

Figure 61: HFR curves for the in-house electrode MEAs. ..................................................................... 63 

Figure 62: Polarization curves for the in-house electrode MEAs. ........................................................ 64 

Figure 63: Normalized polarization curves for the in-house electrode MEAs. ..................................... 65 

 

  



List of Tables 
Table 1: Color designations for rating the transfer result. ................................................................... 33 

Table 2: Description of the experiments for finding the parameter space for the decal transfer 

process. ................................................................................................................................................. 35 

Table 3: Average and standard deviation of the HFR and EIS resistance for the five statistical repeats 

at 1.0 A/cm2. ......................................................................................................................................... 45 

Table 4: Average and standard deviation of the polarization curve voltage for the five statistical 

repeats at 1.0 A/cm2. ............................................................................................................................ 46 

Table 5: Average and standard deviation of the HFR for the five series stacking at 1.0 A/cm2. .......... 49 

 

  



Abbreviations 
 

CCM Catalyst coated membrane 

DSC Differential scanning calorimetry 

ECSA Effective catalyst surface area 

EIS Electrochemical impedance spectroscopy 

FC Fuel cell 

GDL Gas diffusion layer 

HC Hydrocarbon 

HFR High frequency resistance 

LSC Long side chain 

MEA Membrane electrode assembly 

MPL Microporous layer 

MSC Medium side chain 

OCV Open circuit voltage 

PEM Proton exchange membrane / polymer electrolyte membrane 

PFSA Perfluorosulfonic acid 

SSC Short side chain 

Tg Glass transition temperature 

  



Symbols and Operators 
 

Symbol Description Units 

𝑄̇ Heat flux W 

𝑞̇ Specific heat flux W/m2 

𝑚 Mass kg 

𝑐𝑝 Specific heat capacity J/kgK 

𝑇 Temperature K, °C 

A Area m2 

𝑅𝑡 Thermal resistance W/K 

ki Thermal conductivity of a material W/mK 

l Length m 

∆ Difference operator, eg. ∆𝑇  

I/Pt Ratio of ionomer to (platinum) catalyst  

C/Pt Ratio of carbon to (platinum catalyst  

σ Standard deviation  

 

  



Contents 
Abstract ................................................................................................................................................... 3 

Acknowledgements ................................................................................................................................. 4 

List of Figures .......................................................................................................................................... 5 

List of Tables ........................................................................................................................................... 7 

Abbreviations .......................................................................................................................................... 8 

Symbols and Operators ........................................................................................................................... 9 

1 Introduction .................................................................................................................................... 1 

2 Background ..................................................................................................................................... 3 

2.1 Proton Exchange Membrane Fuel Cells (PEM FC)................................................................... 3 

2.2 The Membrane Electrode Assembly (MEA) ............................................................................ 4 

2.3 Significance to PowerCell ........................................................................................................ 5 

2.4 Analysis Methods and Conditions ................................................................................................... 6 

2.4.1 Visual Inspection ................................................................................................................. 6 

2.4.2 Differential Scanning Calorimetry ....................................................................................... 6 

2.4.3 In-Situ Testing Conditions ................................................................................................... 7 

2.4.4 Conditioning Data ............................................................................................................... 8 

2.4.5 Polarization Curve ............................................................................................................... 9 

2.4.6 High Frequency Resistance (HFR) and Electrochemical Impedance Spectroscopy (EIS) .. 10 

2.5 Materials ....................................................................................................................................... 12 

2.5.1 Proton Exchange Membrane (PEM) .................................................................................. 12 

2.5.2 Electrodes.......................................................................................................................... 13 

2.5.2.1 In-House Electrodes .......................................................................................................... 14 

2.5.2.2 Supplier Electrodes ........................................................................................................... 15 

2.5.3 Sub-gaskets ....................................................................................................................... 16 

2.5.4 Gas Diffusion Layer ........................................................................................................... 16 

2.5.5 Decal Substrates ................................................................................................................ 16 

2.5.6 Pressure Distribution Pads ................................................................................................ 16 

3 Glass Transition Temperature ....................................................................................................... 17 

3.1 Initial Experiments ................................................................................................................ 17 

3.2 Discussion .............................................................................................................................. 18 

4 The Decal Transfer Process ........................................................................................................... 20 

4.1 Assembly Plates .................................................................................................................... 20 

4.2 Pressure Distribution and Fitting the Components .............................................................. 22 

4.3 Temperature within the MEA ............................................................................................... 24 

4.4 Series stacking ....................................................................................................................... 25 



5 Electrodes: Quasi-Annealing Behavior .......................................................................................... 29 

5.1 Initial Experiments ................................................................................................................ 29 

5.2 Discussion .............................................................................................................................. 31 

6 Processing Parameter Matrix: What Works? ................................................................................ 32 

6.1 Experiment Description ........................................................................................................ 32 

6.2 Results ................................................................................................................................... 36 

6.3 Discussion .............................................................................................................................. 39 

6.4 Conclusion ............................................................................................................................. 41 

7 Performance of the MEAs ............................................................................................................. 42 

7.1 Experiment Description ........................................................................................................ 42 

7.2 Statistical Evaluation: Results and Discussion ...................................................................... 44 

7.3 Series Stack Performance: Results and Discussion ............................................................... 47 

7.4 Processing parameters: Conditioning Curve Screening ........................................................ 50 

7.5 Varying Materials: Results and Discussion ............................................................................ 58 

7.6 In-house Electrodes: Results and Discussion ........................................................................ 61 

8 Conclusion ..................................................................................................................................... 66 

9 Outlook ......................................................................................................................................... 67 

Bibliography .......................................................................................................................................... 70 

 

 





 

1 
 

1 Introduction 
 

Climate change stands as one of the greatest challenges facing human civilization in the 21st century. 

According to the Intergovernmental Panel on Climate Change (IPCC)’s sixth assessment report, human 

activities have already caused a 1.1 °C increase in global surface temperatures from pre-20th-century 

levels, emphasizing the urgent need to secure a livable and sustainable future [1]. Unsustainable 

energy use, a key driver of climate change, continues to persist, with energy demand projected to 

grow as economies develop [2]. Additionally, the finite nature of fossil fuel resources necessitates the 

adoption of alternative technologies well before depletion occurs. Hence, renewable energy sources 

play a critical role not only in preserving favorable climatic conditions but also in enabling future 

economic prosperity. Efficient energy storage solutions are essential due to the intermittent nature of 

many renewables, and hydrogen holds significant promise in this regard. It can be produced through 

electrolysis, stored, and converted back to energy via fuel cells, offering advantages in mobility, 

scalability, and grid regulation potential. 

The concept of a hydrogen-powered future has captivated the imagination for decades. As far back as 

1874, Jules Verne envisioned in his book The Mysterious Island that water could serve as a fuel source, 

with hydrogen and oxygen providing an inexhaustible source of heat and light [3]. Verne's visionary 

ideas about hydrogen as a fuel source are particularly relevant today as we explore the potential of 

hydrogen-powered technologies. Despite this longstanding fascination, the availability and production 

of hydrogen have posed significant challenges. However, with the increasing scale of renewable power 

generation, the feasibility of periods with abundant and inexpensive energy for electrolysis is 

becoming more attainable. 

Proton exchange membrane fuel cells (PEM FCs) have emerged as a promising technology for 

achieving high-efficiency, low-emission energy conversion, particularly in decentralized and mobile 

applications such as vehicles. At the heart of PEM FCs lies the membrane electrode assembly (MEA), 

where the hydrogen oxidation and oxygen reduction reactions occur. The two main methods of MEA 

production are by direct coating of a catalyst ink onto a PEM, and by decal transfer where the 

electrode is coated onto a decal substrate as a carrier and then transferred onto the membrane. This 

thesis aims to address the crucial aspects of MEA production by decal transfer by systematically 

investigating the effects of temperature, pressure, and transfer time on MEA performance. Through 

a comprehensive experimental approach, the transfer quality will be evaluated across a range of 

processing parameters, and in-situ performance tests will be conducted using a 2k experiment matrix. 

By elucidating the impact of these processing parameters, valuable insights can be gained into the 

electrochemical performance of the resulting MEAs. Furthermore, this research explores the influence 

of different proton exchange membranes and in-house produced cathodes on MEA performance and 

processing parameter space. Supplementary experiments will shed light on the effects of these 

materials on the overall performance of PEM FCs. Attention is also devoted to understanding the glass 

transition temperatures of the membranes and the real temperature profiles experienced by the 

MEAs during the transfer process. These investigations provide a foundation for optimizing the 

transfer process and offer insights into the morphological changes occurring within the electrode. 

Chapter 2 will discuss the background of this thesis, including analytic methods and relevant materials. 

Chapter 3 will discuss glass transition temperature measurements made for different membranes. 

Chapter 4 will describe the decal transfer process, including modifications made during this thesis as 

well as a novel ‘series stacking’ approach which promises greatly increased production capacity. 

Chapter 5 discusses the temperature driven annealing behavior of electrodes observed during the 



2 
 

decal transfer. Chapter 6 describes the observed set of parameters available for a complete transfer 

to occur for different materials. Chapter 7 shows and discussed the in-situ performance 

measurements made in this thesis for MEAs produced with different processing parameters and 

materials. Chapter 8 provides a conclusion to the thesis, and chapter 9 provides an outlook based on 

the results and conclusions from this thesis and further ideas that merit investigation. 

The findings obtained from this study contribute to advancing MEA production knowledge, specifically 

focusing on the effects of processing parameters and materials on the performance of PEM FCs 

through decal transfer. The insights gained from exploring the parameter space, investigating glass 

transition temperatures, temperature profiles, and morphological changes hold the potential to 

inform future optimization strategies. As this research progresses, advancements in catalyst and MEA 

production may be achieved, potentially enhancing PowerCell's in-house capabilities. Furthermore, 

these findings have broader implications for the academic and renewable energy communities, 

offering valuable insights towards the development of more efficient and sustainable energy 

solutions. 

 

  



 

3 
 

2 Background 
 

This section describes in brief the technical background of fuel cells and membrane electrode 

assemblies, as well as the situation at PowerCell that is the impetus for this thesis. 

 

2.1 Proton Exchange Membrane Fuel Cells (PEM FC) 
 

Polymer Electrolyte Membrane Fuel Cells (PEM FC) have long been among the most promising if 

challenging energy systems to be developed. They represent a clean energy storage system with the 

potential for indefinite scalability. While several fuels, including syngas, methanol and biogas, can be 

considered for use in fuel cells, and each presents its own challenges, hydrogen/air is the most widely 

developed fuel source mixture. This includes the research at PowerCell. The chemical reactions in a 

hydrogen fuel cell can be broken down into two half-cell redox reactions: 

Anode (oxidation): 2 𝐻2 → 4 𝐻+ + 4 𝑒−   

Cathode (reduction): 4 𝐻+ + 4 𝑒− +  𝑂2 → 2 𝐻2𝑂   

Figure 1 shows the schematic operation of a hydrogen PEM FC. Hydrogen gas flows via a gas diffusion 

layer to the anode, where the oxidation reaction occurs. Hydrogen ions flow through the polymer 

electrolyte to the cathode, while electrons pass through a load, giving electrical power, before flowing 

to the cathode. Oxygen flows through a gas diffusion layer to the cathode, where together with the 

electrons and hydrogen ions it forms water, which flows out of the cell. 

 

Figure 1: Schematic diagram of a PEM FC. 

 



4 
 

2.2 The Membrane Electrode Assembly (MEA) 
 

The heart of the fuel cell is the membrane electrode assembly (MEA). This is where the cell reactions 

and mass transport all take place in a fuel cell, and an MEA design needs to carefully balance the needs 

of mass transport to the electrodes, mass transport to the membrane and within the membrane, 

electrical conductivity to the electrodes, electrical resistance between the electrodes, catalyst 

availability, degradation resistance, mechanical stability and so on. 

In this work, an MEA refers to an anode, membrane and cathode assembly. This is a 3-layer MEA, as 

opposed to a 5-layer MEA which has additional sub gaskets, or a 7-layer MEA which also has gas 

diffusion layers added on either side (Figure 2). 

 

Figure 2: Schematic overview of 3-, 5- and 7-layer MEAs. 

The MEAs at PowerCell are prepared using a decal transfer procedure. In this method, electrodes are 

prepared by spreading an electrode ink onto a decal substrate, which functions as an easily removable 

carrier material. The anode, membrane and cathode are then laminated together under controlled 

temperature, pressure and time. This method is one of the most common for preparing MEAs and is 

particularly interesting for its potential to be scaled up as a roll-to-roll process. Another major method 

of MEA fabrication is to directly coat catalyst ink onto the membrane; however, that will not be 

focused on in this thesis. 

 

 

 

 

 



 

5 
 

2.3 Significance to PowerCell 
 

PowerCell wishes to produce the MEAs needed for the company’s fuel cell products in-house via a 

decal transfer hot pressing method. The company primarily intends to use in-house electrodes for this 

process. Previous work at PowerCell by S. A. Ogu-Egege [4] has established a general method for 

producing these MEAs, but little work has been done to explore the process parameters, particularly 

pressure, temperature and time in the hot-pressing stage. In addition, it is critical that proper design 

of the hot-pressing step is understood for different materials. New membrane ionomers such as 

hydrocarbon PEMs present new opportunities for performance and ‘tunability’ that need to be 

explored, as well as presenting the chance to move away from fluorinated ionomers. This requires 

increased understanding of the MEA assembly procedure. Furthermore, the company aims to produce 

fuel cells for the aviation sector, which requires the use of catalysts with very high loadings. Exploring 

the assembly process for such catalysts is therefore of enormous and immediate importance to 

PowerCell. 

  



6 
 

2.4 Analysis Methods and Conditions 
 

It is important to describe what means of evaluation are used and what information they can deliver. 

Different means of evaluation have different strengths and different purposes, ranging from quick and 

simple methods to more in-depth methods capable of providing large amounts of detailed 

information. 

 

2.4.1 Visual Inspection 
 

A direct, visual inspection is among the most useful and undervalued analysis methods. It is 

particularly useful as a means of quality control and as a preliminary observation of results. It is notably 

useful for inspecting electrodes, especially in conjunction with backlighting via a light board, to 

determine the quality and consistency of the coating as well as the location of defects and damaged 

areas where a coating is not useable. A visual inspection of an assembled MEA is useful for identifying 

defects such as pinholes where light can shine through, and the removed decals can be analyzed to 

ascertain the completeness of the transfer. When laminating sub gaskets onto a 3-layer MEA, a visual 

inspection allows for the early detection of leaks that may make the MEA unusable. In chapter 6, visual 

inspection is the primary means by which the quality of the transfer is evaluated. 

 

2.4.2 Differential Scanning Calorimetry 
 

Differential Scanning Calorimetry (DSC) is done by slowly heating a small sample of a substance and 

measuring the heat flux through the sample as a function of temperature, relative to a known 

reference. Here, DSC is performed with samples of 10 ±1 mg in sealed aluminum pans under a nitrogen 

atmosphere at a temperature increase rate of 10 K/min, in order to evaluate the glass transition 

temperatures of polymer electrolyte membranes. The glass transition temperature can be seen 

graphically in the DSC by way of a diagram of specific heat flux versus temperature, where the curve 

experiences a decrease [5]. In this thesis, the midpoint glass transition temperature is used. An 

example of a DSC curve is shown in Figure 3. 



 

7 
 

 

Figure 3: Example of a DSC graph for measuring the glass transition temperature of a membrane. 

 

2.4.3 In-Situ Testing Conditions 
 

The resistance properties of the MEA are measured in-situ using a single cell testing approach. An MEA 

is prepared and placed into the testing setup. Figure 4 shows the circuit diagram for in-situ testing. 

The setup is current controlled at the load bank, and the voltage is measured directly from the bipolar 

plates. The hydrogen and air flow within the cell is in excess of the reaction stoichiometry. In-situ tests 

in this thesis are done as 12 cm2 subscale tests under normal operating conditions (NOC) with a cell 

temperature of 65 °C and 60 % relative humidity. The reactant inflow is in excess of the required 

reaction stoichiometry by a factor of 12.5 and 15 at the anode and cathode respectively.  

 

Figure 4: General circuit diagram for in-situ subscale testing. 

 

  

    air      air

 oad 
 ank

 

   

 ell voltage    

E  



8 
 

2.4.4 Conditioning Data 
 

Before in-situ testing, the MEA must be conditioned, as the membrane and catalyst ionomer is not 

fully hydrated. This is done by operating the fuel cell for up to 12 hours under a constant current of 

1.5 A/cm2 at a high relative humidity of 80%, during which the water generated by the reaction 

activates the ionomer. The resulting voltage graph gives information on the water uptake of the 

membrane from the slope of the voltage curve in the beginning once constant current has been 

reached. Some information on the MEA interface can be read from the final, stable voltage and HFR. 

Should these values not reach a constant value, this can indicate a malfunction of the MEA or a minor 

leak in the membrane. The conditioning data is especially useful for such troubleshooting.  

Figure 5 shows the conditioning data for a well-functioning MEA, while Figure 6 shows the conditioning 

data for a poorly functioning MEA. The poorly functioning MEA shows a decreasing voltage and an 

increasing resistance while the well-functioning MEA shows the opposite; moreover the poorly 

functioning MEA also fails to reach a constant steady value. It is likely that a fiber from the GDL or 

something similar caused damage to the membrane. 

 

Figure 5: Example of conditioning data for a well-functioning MEA, from MEA D10 manufactured under T=200C and p=300 
N/cm2 for t=300 s 



 

9 
 

 

Figure 6: Example of conditioning data for a poorly functioning MEA, from MEA D12 manufactured under T=160C and 
p=300 N/cm2 for t=300 s. 

 

2.4.5 Polarization Curve 
 

The polarization curve of a fuel cell is obtained by measuring the voltage of the cell while varying the 

current density. It can be divided into three regions, where kinetic, ohmic and mass transport effects 

are respectively dominant. Also significant is the open circuit voltage (OCV) of the fuel cell. A 

polarization curve with the kinetic, ohmic and mass transport dominated regions marked is shown in 

Figure 7. 

 

Figure 7: Example of a polarization curve with the OCV and the kinetic, ohmic and mass transport regions marked. 

OCV 

Full 

Load 

Kinetic Ohmic Mass Transport 



10 
 

The theoretical OCV of a fuel cell, given from the electrochemical potentials of hydrogen and oxygen, 

is 1.23 V; however, since no fuel cell is perfectly tight and some hydrogen crossover diffusion always 

occurs, this value cannot be reached in practice and the OCV of a well performing fuel cell using 

hydrogen and air as fuel is generally around 0.9 to 1.0 V. An     below  .9 indicates a ‘tightness’ issue 

in the MEA with a greater amount of fuel crossover. An OCV below 0.8 generally indicates a hole in 

the membrane. 

In the kinetic region of the fuel cell, losses are dominated by reaction kinetics. Specifically, the oxygen 

reduction at the cathode is dominant, as the hydrogen oxidation reaction at the anode is very fast. 

This is also the reason why the cathode generally has a higher catalyst loading than the anode in a fuel 

cell.  f particular interest is the ‘kinetic point’ at  .  A/cm2, as this provides information on the 

reaction kinetics and the effective catalyst surface area (ECSA). The main information here is about 

the catalyst behavior rather than the interfaces between PEM and electrodes. 

The losses in the ohmic region are dominated by resistance effects to the flow of charged species 

during the reaction. Generally, the resistance to the flow of electrons is negligible compared to the 

resistance to the flow of protons due to the large difference in mass and mobility between the two. It 

is thus the ionic conductivity of the protons in the electrodes and membrane as well the resistance at 

the interfaces between the membrane and the electrodes that are significant here, meaning that the 

PEM and interface dominate this region. If the PEM is the same between two tests, then the interface 

dominates any differences, and the interface is determined by the decal transfer conditions. To 

analyze the ohmic region, polarization curves can be normalized in the kinetic point, i.e. shifted so 

that the kinetic points overlap. The quasi-linear slope in the ohmic region is of particular interest as it 

provides information about the cell resistance, which is strongly affected by the MEA transfer 

conditions. 

In the mass transport region, the current density is so high that mass transport within the cell becomes 

the limiting factor. It is common in this region that not enough reactant can be supplied to the 

electrodes – mainly to the cathode, since hydrogen diffusion is very fast - through normal background 

diffusion from the bulk gas phase. This results in a concentration gradient as the transport of the 

reactant species becomes a significant source of resistance. When comparing the mass transport 

regions of two polarization curves, the degree of divergence is of note alongside the full-load point at 

2.5 A/cm2. For a well-functioning MEA, this point should be at or above 0.6 V. 

 

2.4.6 High Frequency Resistance (HFR) and Electrochemical Impedance Spectroscopy 

(EIS) 
 

High frequency resistance is measured by applying a perturbation current of at least 1 kHz to the fuel 

cell and measuring the resulting resistance. Because many resistance sources are frequency 

dependent [6], the HFR value ignores most resistance sources and (in the ohmic region) mainly shows 

the ion conduction resistance through the electrodes, membrane and at the interfaces [7]. HFR is 

particularly useful because it can be performed continuously throughout the testing without 

interfering in other measurements or operations. It can also provide information about the hydration 

of the membrane during conditioning, as the hydration state of the membrane is a significant 

contributor to the ionic resistance in the PEM. 

Electrochemical impedance spectroscopy (EIS) obtains a complex resistance from the fuel cell by 

inducing a sinusoidal perturbation and measuring the phase shift delay of the response [8]. This 



 

11 
 

complex resistance is measured across a spectrum of different frequencies, forming a Nyquist plot 

(Figure 8). In this thesis, the real component of the EIS is desired, and so EIS resistance values are 

taken as the Y-axis intercept of the Nyquist plot. This is considered a very high-accuracy method of 

obtaining the internal resistance of the fuel cell, as any remaining complex portion of the resistance 

that may remain in the HFR is resolved. 

 

Figure 8: A Nyquist plot, obtained by measuring the complex EIS resistance between 5000 and 100 Hz. 

In this thesis, two methods of finding the resistance are employed: EIS, and a steady-HFR, which takes 

the resistance at a 1 kHz steady perturbation frequency. While the steady HFR and EIS are both high-

frequency resistance methods, in this thesis references to “   ” refer to the steady   k z     

specifically while “E  ” values refer to the real portion of a complex EIS measurement between 5000 

and 100 Hz. 

Note also that while the EIS value is more exact due to being able to eliminate the imaginary 

component of the Nyquist plot entirely, it cannot be measured at the same time as the polarization 

curve. Because of this, the MEA has a different testing history when the EIS is measured compared to 

when the HFR and polarization curves are measured, which can affect results. 

  



12 
 

2.5 Materials 
This chapter describes the different materials relevant to the production of MEAs by the decal transfer 

process. This includes the components of the MEA, the electrodes, the proton exchange membrane, 

the sub gaskets and the gas diffusion layer, as well as materials in the assembly process that affect the 

transfer, the pressure distribution pads and the decal substrate. 

 

Figure 9: Schematic overview of the materials in the MEA, as well as supplementary materials for the decal transfer. 

 

2.5.1 Proton Exchange Membrane (PEM) 
 

The proton exchange membrane separates the two electrodes, and thereby the two half-cell 

reactions. It must be electrically insulating, yet conductive to protons. To achieve this, the membrane 

is composed of a thin sheet of ionomer, which is electrically insulating, yet forms conductive channels 

from polar −𝑆𝑂3𝐻 groups. Most state-of-the-art PEMs have a thickness between 10 and 20 μm. 

An ionomer is a thermoplastic polymer which is stabilized by ionic cross-linkages. Generally, an 

ionomer has a main backbone polymer chain, and various side chains branching off from it which 

contain the ionic group (−𝑆𝑂3𝐻). The structure of an ionomer is shown in Figure 10. 



 

13 
 

 

Figure 10: Structure of an Ionomer (Nafion). 

The hydrophilic ion groups tend to orient themselves together, which on a very small thickness scale 

creates channels with inward facing ion groups. These ion groups attract water due to their polarity, 

and as they are acidic form a highly protonated electrolyte within the channels. Whenever a new 

proton is introduced on one side of the channel, another proton is ejected from the other side of the 

channel to maintain charge neutrality, and by this method the channel can be used to conduct protons 

(or other cations). 

Most membranes use a perfluoro sulfonic acid (PFSA) ionomer, with hydrocarbon (HC) PEMs more 

recently emerging into focus for commercial application. In this work, four membranes are studied. 

PFSA ionomers are broadly categorized by the length of their side chain. 

Membrane A is a PFSA membrane with a medium side chain (MSC) length that is 12 microns thick. 

Membrane B is a PFSA membrane with a long side chain (LSC) length that is 15 microns thick. 

Membrane C is a PFSA membrane with a short side chain (SSC) length that is 12 microns thick. 

Membrane D is a HC membrane that is 7 microns thick. 

Short side chain ionomers offer a lower proton conductivity resistance than long side chain ionomers 

as the density of ion groups in the structure is greater. 

 

2.5.2 Electrodes 
 

Within this thesis, a fuel cell electrode consists of a catalyst (Pt on carbon) in a highly dispersed form 

mixed with an ionomer. The reaction takes place at a 3-phase boundary, where the catalyst has access 

to an electrical connection (via the carbon support) for the electrons to flow to/from, and an ionic 

connection (through the ionomer) for protons to flow. This is the case for both anode and cathode, 

though the difference in the reaction rates means that the cathode requires a higher catalyst loading. 

In this thesis, electrodes are either sourced from Supplier A or produced in-house. 

 



14 
 

2.5.2.1  In-House Electrodes 
 

In general, to prepare an electrode ink, the catalyst (Pt on carbon) is mixed with ionomer and a solvent 

consisting of water and various alcohols. This is left to mix overnight, either by use of ceramic milling 

balls or a stirring bar. 

Then, the mixture is dispersed via ultrasonic dispersion. Depending on the recipe, this may occur in 

multiple steps with additional stirring in between. Once this catalyst ink is finished, rheological 

measurements are taken for quality control. 

A decal substrate material is selected, and a doctor blade is used to distribute the ink over the decal 

substrate. Once the decal has been coated, it is left to dry under its own atmosphere. Then, a quality 

control of the electrodes is carried out. Here the electrodes are examined using a light board to assess 

the evenness of the coating and under a microscope to assess whether the electrodes show significant 

cracks or agglomerates. If the electrode passes muster, it is ready for the MEA transfer process. The 

electrode preparation follows the work of E. Ulberstad in the context of her master’s thesis [9]. 

 

Figure 11: An In-house prepared cathode, visual inspection using a light board. 

In this thesis, the in-house electrode materials were supplied by the catalyst lab with an I/Pt ionomer 

to catalyst ratio of 1.0 in each case. The inks were produced from catalyst material (high or low surface 

area Pt/C) with a carbon support to catalyst ratio of 1.0, a dry content of 12.6 % by mass, and a 60:40 

solvent matrix of water and 1-propanol. 

Cathode 1 used a high surface area Pt/C catalyst and ionomer 1. The ink was dispersed up to a total 

energy of 200 kWs in 50 kWs cycles interspersed by 30 minute resting periods. The average loading of 

the produced decals was measured at 0.262 mgPt/cm2 with a standard deviation of 0.014 mgPt/cm2 

across six measurements with a sample size of 25 cm2. 

Anode 1 used a low surface area Pt/C catalyst and ionomer 2. The ink was dispersed up to a total 

energy of 200 kWs in 50 kWs cycles interspersed by 30 minute resting periods. The given loading of 

the electrode was 0.1 mgPt/cm2. 

Cathode 2 used a high surface area Pt/C catalyst and ionomer 2. The ink was dispersed up to a total 

energy of 200 kWs in 50 kWs cycles interspersed by 30 minute resting periods. The given loading of 

the electrode was 0.25 mgPt/cm2. 



 

15 
 

Cathode 3 used a high surface area Pt/C catalyst and ionomer 2. The ink was dispersed up to a total 

energy of 50 kWs. The given loading of the electrode was 0.25 mgPt/cm2. 

Cathode 4 used a high surface area Pt/C catalyst and ionomer 2. The ink was dispersed up to a total 

energy of 10 kWs. The given loading of the electrode was 0.25 mgPt/cm2. 

 

2.5.2.2  Supplier Electrodes 
 

All externally supplied electrodes came from Supplier A. Two batches each of anode and cathode were 

available, and each pair had the same specifications. The anode electrodes had an I/Pt ionomer to 

catalyst ratio of 0.7, with a target loading of 0.05 and a measured loading of 0.06 mgPt/cm2 for each 

batch. The cathode electrodes had an I/Pt ratio of 0.9, with a target loading of 0.25 and a measured 

loading of 0.24 and 0.27 mgPt/cm2 respectively for the two batches. In each case, the C/Pt carbon 

support to catalyst ratio was 1.0. 

At outset of the thesis, a thorough visual inspection of the supplier electrodes was conducted. This 

showed several issues that had to be considered when using the electrodes. These issues were most 

visible when backlit by a light board, and include pinholes, dry regions, coating waves and smudges 

(see Figure 12). Such problem areas were marked and avoided when selecting material to work with. 

 

Figure 12: Supplier A electrode decals with visible defects. 

Coating waves and dry regions are problems that stem from the coating process. Pinholes and 

smudges are more likely to occur during the storage and transportation of decals, though pinholes can 

also be a result of an incomplete dispersion or a local agglomeration of the catalyst particles. Prior to 

hot pressing, during which the ionomer in the electrodes helps bind the material together similar to 

an annealing process, the electrodes are extremely fragile, and at present no good solution to their 

storage and transport is available. Fortunately, these issues are easily detectable via visual inspection. 



16 
 

Other parameters, such as agglomeration or dispersion, are harder to detect visually; this is a major 

research focus for the catalyst lab. 

Even while avoiding the problematic areas, some variance in quality was noted. This variance was 

present between different batches and different electrode sheets within each batch. Therefore, the 

Mx experiments mix batches in the four-cell array, such that each batch of cathode has one sample 

with each batch of anode. In chapter 7 resistance measurements are carried out on MEAs prepared 

from different sheets and batches in order to establish a statistical error margin for electrical 

measurements. A significant part of this variance likely stems from these differences in electrode 

quality. 

 

2.5.3 Sub-gaskets 
 

The sub-gaskets consist of a polymer base film and a heat activated adhesive. They are mounted on 

the MEA such that the adhesive side is facing inwards and a window with a defined area is left free for 

the reaction to take place. Their function is to enable the mounting of the MEA into a testing cell 

without any leaks across the cell. The assembly of the sub-gasket is a lamination procedure which 

requires moderate heat and pressure applied using an appropriate hand tool. 

 

2.5.4 Gas Diffusion Layer 
 

The gas diffusion layer is a thin layer of a porous material which serves the purpose of enabling an 

even spread of the gases across the electrodes. In this thesis, carbon paper is used for this purpose. 

 

2.5.5 Decal Substrates 
 

The decal substrate is the carrier material for the electrodes before and during the decal transfer. They 

must be chemically inert, thermally stable across the operating conditions of the hot press and have 

excellent release properties. Here, glass fiber reinforced Teflon is used as a decal substrate. 

 

2.5.6 Pressure Distribution Pads 
 

The use of pressure distribution pads is a recent addition to the MEA assembly procedure at 

PowerCell, introduced by S. A. Ogu-Egege [4]. At present, disposable pressure distribution pads made 

of cellulose are used. Their function is to ensure an even distribution of pressure across the MEA 

during the transfer process. 

  



 

17 
 

3 Glass Transition Temperature 
 

A phenomenon of amorphous polymers, including ionomers, is the glass transition temperature, 

where the material undergoes a change from a rigid glassy state to a more malleable rubbery state. 

This change comes with significant differences in the physical and mechanical properties of the 

ionomer. In the decal transfer process, the malleable state is significant for the transfer behavior and 

MEA properties. 

While an investigation of the glass transition behavior of ionomers is not the purpose of this thesis, it 

is important to consider the glass transition temperature of the membranes in order to understand 

the effects of temperature on both transferability and performance of the MEA in other experiments. 

Thus, glass transition temperature measurements are conducted using DSC. In future work, a more in-

depth study of the glass transition behavior may be needed; these measurements are not repeated, 

are only performed on the ionomers in the membranes A, B, C and D, and are only performed in one 

direction, i.e., heating up the sample rather than cooling it down. The mechanical properties of the 

ionomers above and below the glass transition temperature are also not measured. These 

measurements should serve to support later experiments by giving context to the glass transition 

temperatures of relevant materials. 

 

3.1 Initial Experiments 
 

The glass transition temperatures of membranes A, B, C and D are measured by DSC as described in 

section 2.4.2. Figure 13 shows the DSC graphs for membranes A, B, C an D. 



18 
 

 

Figure 13: DSC results for membranes A, B, C and D. 

Membrane A shows a (midpoint) glass transition temperature at 157 °C, membrane B at 162 °C, and 

membrane D, which is a hydrocarbon ionomer, shows a glass transition at 188 °C. Membrane C shows 

no glass transition. Most likely the sample was not compressed sufficiently in the pan, leading to poor 

contact with the aluminum and obscuring the specific heat transport relative to the (empty) reference 

pan. 

It is important to note that these measurements are single measurements conducted in only one 

direction (heating the sample, not cooling the sample). They are intended to provide general values 

for the glass transition temperature to inform the discussion around other results. There are many 

effects and observations around the glass transition, such as the different observed shapes of the DSC 

graphs and the annealing behavior described in chapter 5, which warrant further investigation than is 

possible in this thesis. 

 

3.2 Discussion 
 

While the Tg measurement failed for membrane C, the results show a Tg at 157 and 162 °C respectively 

for membranes A and B. Such a low difference is expected, as both are PFSA membranes which mainly 

deviate in the length and frequency of their side chains, and it is likely that a successful measurement 

of membrane C would also show a Tg near 160 °C. Membrane D showed a significantly higher Tg at 

188 °C, which correlates to the significant difference of a HC ionomer to a PFSA ionomer. 



 

19 
 

While this is a sufficient investigation of the glass transition temperature for this thesis, there remain 

many questions. The shape of the graphs varies significantly, and the greatest variance is between the 

two successful tests of PFSA membranes (A and B), which are expected to have similar structures and 

morphology. It may also be worthwhile, in future work, to test the degradation temperature of the 

ionomers. In this thesis, the glass transition is mainly interesting as context for the temperature-driven 

behavior in other tests, so these measurements should be sufficient. Further work may need to 

investigate the glass transition in greater detail.  



20 
 

4 The Decal Transfer Process 
 

At PowerCell, the membrane electrode assembly is produced using a decal transfer process. In this 

method the anode, the membrane and the cathode are laminated into an MEA at elevated 

temperature and pressure in a single processing step. The electrodes are provided on a decal 

substrate, a chemically inert carrier material with good release properties and high thermal stability, 

and this process transfers them onto the membrane while also ‘annealing’ each electrode from a 

homogenized mix of ionomer and catalyst powder into a solid material. 

The decal transfer process is done using a programmable hot press. The MEA sample is placed on a 

suspended steel tray within the press where it remains until the top and bottom plates of the press 

have reached their desired temperature. The bottom plate then closes towards the top plate, lifting 

the suspended tray and sample in the process. The pressing commences for the defined time at the 

defined pressure, following which the top and bottom plates open, and cooling cassettes are used for 

120 seconds to return the sample and tray to room temperature. 

At the outset of work on this thesis, a functional operating point for the successful transfer of the 

electrodes was known from previous work at PowerCell [4] at 160 °C pressing temperature, 300 

seconds of pressing time, at a pressure of 300 N/cm2 (operating point D4 in this thesis). The same work 

introduced pressure distribution pads consisting of cellulose in order to ensure an even distribution 

of pressure within the MEA. 

During this thesis, the following alterations were made to the basic process: 

- ‘ itting’ the cellulose pressure distribution pads  membrane and electrodes to the same size to better 

control the distribution of pressure 

- Introducing assembly plates to streamline MEA preparation outside the press and allow for the 

simultaneous preparation of multiple samples 

Furthermore, the temperatures within the MEA during pressing were examined, as opposed to the 

temperatures of the top and bottom plates of the press, and the possibility of pressing multiple MEAs 

simultaneously in series as a stack was explored. 

 

4.1 Assembly Plates 
 

Assembly plates were introduced to the process primarily in order to facilitate the rapid production 

of MEA samples of varying sizes in large quantities, as is necessary for the ‘Mx matrix’ form of 

experiments used in this thesis. The assembly plates consist of two flat steel plates, one of which has 

placement markings on it, upon which samples are arranged. It is then possible to arrange samples 

outside of the hot press and, by switching between different sets of plates, maximize the use time of 

the hot press. Sample arrangement is then far simpler and can be accomplished while the previous 

MEA is being produced. Figure 14 shows two sets of assembly plates with samples. 



 

21 
 

 

Figure 14: Assembly plates. 

Beyond simplifying the sample arrangement and increasing the efficiency with which the press can be 

used, these plates also avoid a problem whereby slightly misaligned samples may have sections of 

membrane ‘poke out’ from underneath the cellulose and can stick to the hot press. Further, they can 

serve to protect the hot press from damage, as was the case during an attempt to measure the internal 

temperature of the MEA during pressing in absence of the cellulose pressure distribution pads (Figure 

15). 

 

Figure 15: Assembly plate damaged instead of the hot press. 

In future work at PowerCell, it may be feasible to expand this concept to accommodate larger MEA 

samples needed for full scale prototyping or commercial production, possibly including integration 

with the current suspended tray setup. 

 

 

 



22 
 

4.2 Pressure Distribution and Fitting the Components 
 

As determined by S. A. Ogu-Egege [4], a pressure distribution pad is required to ensure an even 

distribution of pressure across the MEA during the transfer, and single use cellulose pads provide a 

good and economical solution. With the size of this pad exceeding the size of the membrane, which 

in turn exceeds the size of the electrodes, it is necessary to better understand the distribution of 

pressure between areas of the sample with different materials. 

An experiment was set up in which a sample was pressed together with a layer of pressure sensitive 

film (FujiFilm Prescale LW). Figure 16 shows the general setup of the experiment. 

 

Figure 16: Pressure distribution measurement general setup. 

Between the assembly plates is first a layer of the cellulose pressure distribution pad, upon which is 

the MEA, upon which is the pressure sensitive film. The experiment was carried out using cellulose 

pads with dimensions of 7 x 7 cm, a commensurate area for the pressure film, and four MEA samples 

with dimensions of 2 x 2 cm arranged in a grid pattern. The experiment was run at a pressure of 300 

N/cm2 relative to the area of the cellulose, a pressing time of 300 s, and a temperature of 30 °C to 

accommodate the pressure sensitive film. The MEA used Supplier A anode and cathode material and 

membrane A. Figure 17 shows the results of the experiment. 

 

Figure 17: Initial prescale pressing test including cellulose pressure distribution layers, with four 2x2 MEA samples, showing 
areas with and without electrodes. 

Clearly visible is an imbalance between areas of the cellulose with and without MEA, with the areas 

with the MEA seeing a higher pressure. The distribution of this pressure is very even for each MEA and 

between the four samples. However, the actual pressure upon the MEA remains unknown, which 

somewhat complicates efforts to characterize and control the effects of pressure on the transfer 

process. 



 

23 
 

Interestingly, the edge of the electrode experiences greater pressure than the rest of the electrode. It 

may be that when the electrode is cut, the blade pushes electrode or decal material to the side, 

creating a localized increase in the height of the electrode. Fortunately, the change appears to be too 

localized to be a significant contributor to MEA performance. 

A ‘fitting’ approach is used to resolve the pressure distribution problem.  n this method  cutting die 

are used to prepare membranes, electrodes and cellulose pads such that all possess the same 

dimensions. This avoids having different ‘zones’ with different actual pressures. 

Testing fitted cellulose with four 2.5x2.5 cells and a 33 cm^2 configuration that can be cut down for 

the BFC 12 gives the following results: 

 

Figure 18: pressure distribution in the hot press for the four-cell array. 

 

Figure 19: Pressure distribution in the hot press for a 33 cm^2 sample. 

In both images, there is an even pressure distribution, which shows the viability of the fitted cellulose 

configuration. Further, in Figure 18 we see that the pressure distribution is also even between 

separated samples, which shows the viability of pressing multiple samples simultaneously if their 

materials and the process parameters are the same. Going forward, all samples are prepared using 

cutting die, with electrodes, membrane and cellulose pressure pads all of the same dimensions. 

 

 



24 
 

4.3 Temperature within the MEA 
 

The decal transfer process takes place at an elevated temperature. In previous work [4], 160 °C was 

established as a good operating temperature for achieving a good quality decal transfer with 

Membrane A and both select in-house electrodes and electrodes from Supplier A. The MEA, as well as 

surrounding materials including the decal substrates and pressure distribution pads, must all reach 

this temperature. For this to occur, heat must be supplied. 

𝑄̇ = 𝑚 ∙ 𝑐𝑝 ∙ ∆𝑇 

The heat source is the heated plates of the hot press. For each layer in between these plates and the 

MEA the transfer of heat becomes slower because the heat transport resistance increases. For a series 

of solid layers, this can be described as 

𝑄 = 𝑞̇ ∙ 𝐴 =
𝑇𝑝𝑙𝑎𝑡𝑒−𝑇𝑀𝐸𝐴

𝑅𝑡

̇
  𝑅𝑡 =

𝑙1

𝑘1∙𝐴
+

𝑙2

𝑘2∙𝐴
+ ⋯ 

It is clear, then, that the introduction of additional layers between the MEA and the hot plates of the 

press must increase the thermal resistance of the assembly and that the MEA will therefore take a 

longer time to reach the required temperature. For each layer, this effect increases proportionally to 

the thickness 𝑙𝑖 of the materials and decreases proportionally to the thermal conductivity of the 

requisite material 𝑘𝑖. 

With the introduction of several layers beyond the decal substrate of the electrodes – specifically the 

assembly plates and cellulose pads - it is necessary to characterize the course of the temperature 

within the MEA during the transfer. To this end, an experiment is conducted to measure the 

temperature of the hot plates of the press and the MEA during a transfer. Temperature readings from 

the plates are taken directly from the press, while the internal temperature of the MEA is taken via a 

type T thermocouple added to the sample in between the top electrode decal and cellulose pad. The 

transfer is conducted at the ‘D4’ operating point at  6  °C, 300 N/cm2 and with 300 s of pressing time. 

 

Figure 20: Internal temperature measurement during a decal transfer. 

The temperatures of the hot press plates and the internal temperature of the MEA during the pressing 

are presented in Figure 21. 



 

25 
 

 

Figure 21: Temperature course of the MEA and hot plates during a decal transfer. 

During the warmup of the press, the temperature of the MEA naturally remains constant at room 

temperature. Following the warmup, the press closes. This begins with the lower plate of the press 

lifting up and carrying the suspended tray with the MEA with it. This is why the temperature of the 

lower plate drops before t=0, as heat begins to flow from the lower plate to the suspended tray and 

the sample. The MEA temperature, T probe, begins to increase at this point as well. 

At t=0 s, the hot press is closed; as heat flows from both the upper and lower plates to the sample, 

the temperature of the upper plate is decreased and the temperature of the MEA increases rapidly. 

Shortly thereafter, the temperatures of the plates begin to stabilize as the hot press adjusts their 

internal heating. Meanwhile, the heating of the sample slows down as the temperature difference 

between plate and sample, which is the driving force for the heat transfer, decreases. 

At t=120 s, the internal temperature of the MEA has reached the desired temperature of 160 ± 5 °C, 

and the temperatures of the hot plates have stabilized. The various temperature profiles remain 

steady until t=300 s, when the press opens. The MEA at first cools down slowly, then rapidly as the 

press closes again with cooling cassettes in place. 

These results generally follow expectations. The interesting unknown that they reveal is what manner 

of time characteristic is needed, at minimum, for the internal temperature to approach the set 

temperature. At 120 s, the gap closes to within ca. 5 °C, and continues to increase logarithmically 

towards the set temperature. At t=300 s, the difference between the internal temperature of the MEA 

and the set temperature of the press is negligible. 

 

4.4 Series stacking 
 

In order to support eventual commercial production, it is important for the decal transfer process to 

be upscaled considerably. Decreasing the time seems to have stark limits, and so a better approach 

would be to increase the MEA production volume with each pressing. As demonstrated in section 4.2, 

0

20

40

60

80

100

120

140

160

180

T 
[C

]

t

MEA Internal Temperature Graph

T probe (single MEA) T upper T lower t=0 t=120 t=300



26 
 

the use of pressure distribution pads makes a parallel pressing of multiple samples viable. What about 

stacking multiple MEAs in series? 

Towards this end, it is necessary to verify that the pressure distribution remains even and find the 

temperature curve in the center of such an MEA stack. In these investigations 8 MEAs are used, with 

membrane A electrodes from Supplier A. 

In order to investigate the pressure distribution in the stack, an experiment is set up with pressure 

sensitive film (Fujifilm Prescale LW) every 2 MEAs. The setup can be seen in Figure 22 and is operated 

in the same manner as the pressure distribution experiment in section 4.2. 

 

Figure 22: Pressure distribution experiment setup for multiple MEAs. 

The results of this experiment are shown in Figure 23. To the left is the result while lit from above; to 

the right the result is backlit using a light board. 

 

Figure 23: The pressure distribution during transfer in a stack of multiple MEAs. 



 

27 
 

The pressure distribution is mostly even, with some notable, localized inconsistencies. A closer 

inspection of the electrodes in these locations shows equivalent defects, such that this localized 

difference seems to speak more to the electrode quality than an inherent imbalance in the distribution 

of pressure (Figure 24). The electrode quality varies here because these MEAs were produced from 

rejected electrode material on account of them not being required to undergo resistance testing. 

 

Figure 24: Localized electrode defects appear on the pressure film. 

Aside from these localized defect areas the pressure distribution is even across the MEA, and across 

the stack. This suggests that transferring multiple MEAs simultaneously and in series is, from a 

pressure perspective, viable even without the introduction of further intermediate pressure 

distribution layers. 

It remains necessary to explore the temperature curve inside the MEAs during transfer. With a hot 

plate above and below, the greatest thermal resistance is to the center of the stack. An experiment is 

conducted in the same manner as the temperature curve measurement in section 4.3, but with a stack 

of 8 MEAs rather than one, and with the temperature probe located at the center of the stack, 

between MEAs 4 and 5. The results are shown in Figure 25 alongside the initial temperature curve for 

a single MEA. 

 

Figure 25: Internal temperature course during transfer for one MEA and 8 MEAs. 

0

20

40

60

80

100

120

140

160

180

T 
[C

]

t

MEA Internal Temperature Graph

T probe (single MEA) t probe (8 MEA) T upper T lower t=0 t=120 t=300



28 
 

The temperature curve within the stack of 8 MEAs is qualitatively similar to that of the sample with 

only one MEA. What is apparent is the effects of the increased distance – and therefore thermal 

resistance – on the change in temperature, as the temperature curve is delayed during the heating 

and cooling sections. By t=120 s, the temperature curve of the 8 MEA stack is comparable to the curve 

for one MEA. Together with the pressure distribution, this suggests that a series stacking approach to 

increasing the production volume of MEAs is viable, and that the number of MEAs could be increased 

beyond the 8 used in this experiment, so long as the pressing time is sufficiently long. 

This provides a very promising outlook for future MEA production on a large scale. The hot press is 

large enough to accommodate a small number of commercial scales MEAs in parallel. With 

appropriate logistical preparation, it should be possible to complete and switch the stacks within 15 

minutes. Should four stacks of 8 MEAs be pressed simultaneously, this suggests a possible throughput 

of 128 MEAs per hour, which, if implemented, would be sufficient to sustain commercial production 

at PowerCell for the foreseeable future. 

The in-situ performance is evaluated in chapter 7.3. A stack of 9 MEAs is used for this to maintain 

symmetry around the central MEA. 

  



 

29 
 

5 Electrodes: Quasi-Annealing Behavior 
 

During transfer  the electrode material undergoes a significant change. The ‘raw’ electrode before the 

transfer is more matte in texture and is very vulnerable to physical damage such as being smudged, 

while the ‘annealed’ electrode after the transfer has a shinier texture and feels like a plastic film which 

is much more resistant to physical damage. This behavior is likely due to the ionomer softening and 

binding the catalyst particles together, and takes place for both in-house produced electrodes and the 

electrodes from Supplier A.  

This annealing behavior is a poorly understood effect with significant implications which cannot be 

thoroughly explored in this thesis due to time constraints. Here, some initial tests to explore this 

phenomenon are done, and some of the possible applications, implications, and experiments 

surrounding this behavior are discussed. Future work at PowerCell should explore this phenomenon 

further. 

 

5.1 Initial Experiments 
 

Some initial tests are conducted to observe the annealing behavior. Time constraints limit the amount 

of testing that can be done, so these are preliminary experiments aimed at providing a general 

overview and understanding of the behavior so that a more comprehensive test plan can be 

performed in future work. 

Three tests are used to gain a qualitative understanding of the annealing behavior - ‘observing 

reflection’  ‘smudging’ and ‘wiping’ the electrode. The raw electrode is matte black with very little 

‘reflection’; during annealing the electrode gains a shiny appearance comparable to a black plastic 

film, which can be used to characterize the annealing. The annealed electrode is also much more 

resistant to ‘smudging’  and a qualitative assessment of the normal and shear forces required to 

smudge the electrode is used to characterize the annealing process. A ‘wiping’ test is performed  in 

which a solvent-laced cloth is used to remove the electrode from the decal. This occurs even for a fully 

developed MEA, however the force required is considerably greater for an annealed electrode. 

Three annealed electrodes are prepared. This is done in a similar manner to the transfer method for 

an MEA described in chapter 4 with the difference that instead of an MEA assembly of anode, 

membrane and cathode only an electrode is used, which is covered by a layer of decal substrate on its 

exposed side. In each case, cathode material from supplier A is used. The first sample is prepared at T 

= 160 °C, p = 300 N/cm2, and t = 300 s (p300 sample). From observations in chapter 5 it is expected 

that the annealing behavior is temperature driven rather than pressure driven, and so the second and 

third samples are prepared with the same time and temperature, but at pressures of p = 0 N/cm2 (p0) 

and p = 5 N/cm2 (p5). These points are selected because it may be useful to introduce an annealing 

step during the preparation of in-house electrodes after the drying step so that the electrodes are less 

fragile in storage and in case there may be advantages for layering multiple electrode coatings. In this 

case it may be simpler to use a hot plate with low downforce rather than a hot press. 

The reflexion behavior of the raw, p0 annealed electrode and p300 annealed electrode is shown in 

Figure 26.  



30 
 

 

Figure 26: Reflexion behavior for (a) the raw electrode, (b) the p0 annealed electrode and (c) the p300 annealed electrode. 

The reflexion of the raw electrode is very low, and indeed almost non-existent, the reflection of the 

p300 annealed electrode is comparable to that of well transferred MEAs, and the reflection of the p0 

annealed electrode is in between, certainly present but not nearly as developed. The reflection of the 

p5 annealed electrode was observed to be shinier than the p0, but not as shiny as the p300 annealed 

electrode. 

The raw electrode is very easily smudged with little to no downforce required, which is highly 

problematic for electrode storage. The p0, p5 and p300 electrodes all required far more force to be 

damaged, with increasing pressure leading to improved smudging resistance. Of note is that even with 

no pressure, the smudging resistance was sufficient to prevent most accidental damage, and that the 

difference between the behavior of the p5 and p300 annealed electrodes was comparable to the 

difference between the p0 and p5, suggesting greatly diminishing returns with increasing pressure. 

The wiping test had very similar results to the smudging test, with the difference that even a fully 

transferred MEA is vulnerable to such an attack. In this case, considerable downforce and multiple 

passes are required to remove the electrode. Of note is that for the annealed electrodes, larger 

sections tended to break off together when sufficient force was applied, which confirms the improved 

internal cohesion due to the annealing. 

Upon conclusion of the reflection, smudging and wiping tests, an MEA transfer using annealed 

electrodes was performed. Anode and cathode were annealed with the same protocol as the p300 

annealed electrode using supplier A materials, and membrane A was used. The transfer occurred at 

operating point D4 (T = 160 °C, p = 300 N/cm2, and t = 300 s). This resulted in a successful MEA transfer 

without residue on the decals, as seen in Figure 27, indicating that pre-annealing the electrodes does 

not interfere with the decal transfer behavior. 



 

31 
 

 

Figure 27: MEA transfer result using p300 annealed electrodes. 

 

5.2 Discussion 
 

Temperature seems to be the greatest driver of the observed annealing behavior, with pressure 

playing a supplementary role. Some pressure seems to be advantageous to the annealing process, but 

this may be simply due to the pressure facilitating a better interface for heat transfer in the hot press. 

This could be examined further using heated plates without cellulose pressure distribution pads. The 

effect of time is yet an unknown, though results from chapters 6 and 7 suggest it may be relevant. Of 

great significance is that a pre annealing step does not appear to challenge the transferability of the 

electrode. This could be explored further in an experiment like the matrix experiments in chapter 6. 

For further experiments, it would be useful to establish a quantitative way of assessing annealing 

behavior. 

The observed annealing behavior appears to have a major impact on the morphology and structure of 

the electrodes. It is unclear how the difference in the electrode surface may impact interactions with 

other components, such as the GDL. When pre-annealing the electrode there may be an effect on the 

interface between electrode and membrane which needs to be explored. It may also be possible to 

use such a step to ‘heal’ cracky electrodes  a phenomenon explored by S. A. Ogu-Egege [4]. A pre 

annealing step could be used for such crack healing while being optimized separately from the decal 

transfer step. It may also be of use in the pursuit of electrodes with higher catalyst loadings, as the 

more robust, annealed electrode might lend itself to being produced in multiple layers. Such an 

approach risks creating additional interface effects, but may provide other opportunities, such as 

controlling the gradient of ionomer across the height of an electrode. If different ionomers are used 

in the membrane and electrode, it may be useful to expose the electrode to a higher temperature in 

the pre-annealing step than the membrane can support. It may then be useful to consider the glass 

transition temperatures of the ionomer used in the electrode in future work. 

 

  



32 
 

6 Processing Parameter Matrix: What Works? 
 

In this chapter, an experiment matrix is set up in order to analyze, by visual inspection, the range of 

parameters available to achieve a successful decal transfer for the MEA. This should provide insight 

into the design rules for a successful MEA transfer and inform the usable operating points for later 

experiments. 

The most significant processing parameters for the decal transfer of the MEA are temperature T, 

pressure p and the time t the MEA is exposed to these factors. At outset of work on this thesis, only 

select operating points were known to transfer reliably. By varying these parameters, an experiment 

matrix can be set up to explore the processing parameters available for a complete transfer. 

The completeness of the transfer can be confirmed visually by the residual electrode left on the decal 

substrate after transfer. There is a dependency on the decal substrate used. However, this is not varied 

within the bounds of this thesis. The parameters explored are the set parameters on the hot press; 

for pressure and transfer time, this corresponds directly to the conditions of the MEA; for the 

temperature this is not the case for the entire pressing time, as discussed in section 4.3, which must 

be taken into account during the analysis of this data. 

 

6.1 Experiment Description 
 

Prior to this thesis, the operating point at T = 160 °C, p = 300 N/cm2 and t = 300 s had been used to 

achieve a successful transfer with membrane A and both supplier A and in-house electrodes. This 

operating point was, therefore, used to inform the experiment matrix for this exploration of the 

available processing parameters. 

Five temperature levels were selected, 120, 140, 160, 180 and 200 °C, around the temperature of the 

known operating point. The distance of 20 °C was chosen because the regulatory behavior of the hot 

press results in temperatures of ± 5 °C, and so differences of 20 °C should always be significant.  

Five pressure levels were selected, 50, 150, 300, 450 and 600 N/cm2, around the pressure of the known 

operating point with the lowest pressure adjusted upwards to maintain some compression of the 

MEA. 600 N/cm2 was taken as an upper bound on account of literature showing greatly diminishing 

returns from increasing the pressure beyond this point [10]. 

Two time levels were selected, 120 and 300 s. The upper time is the pressing time used for the known 

operating point; the lower time was selected at the point where the set temperature of the press is 

reached by the MEA within reasonable deviation (see section 4.3). 

The experiment design matrix is shown in Figure 28 below. The cell designations form a useful 

shorthand for their respective processing parameters. The known operating point at T = 160 °C, p = 

300 N/cm2 and t = 300 s is assigned to cell D4 by this method. 



 

33 
 

 

Figure 28: Experiment design matrix for exploring the space of available processing parameters. 

As seen in section 224.2, the pressure distribution remains the same even when multiple samples are 

being transferred in parallel. For these parameter matrix experiments, an array of four 2.5 x 2.5 cm 

samples was used for each test. This was done so that the anode and cathode materials can be varied. 

For each array, material from two sheets of anode and two sheets of cathode was used, and this 

material was distributed between the four cells so that each cathode material was paired with each 

anode material. In this way, some of the variance between different electrode sheets and batches can 

be counteracted, and the array was better able to show whether a transfer behavior was consistent. 

Table 1: Color designations for rating the transfer result. 

Dark Green Successful transfer with no residual 

Light Green Successful transfer with some residual 

Yellow Inconsistent success of transfer, or high residual 

Red Failed transfer 

 

The transfer was observed visually by the residual of the electrode on the decal and given a color 

based rating. Dark green (Figure 29) represents a full, successful transfer on all samples in the array. 

Light green (Figure 30) represents a successful transfer with minor residual of electrode left on the 

decal, either as individual points or in the texture of the decal. Yellow (Figure 31) represents a 

successful transfer with significant residual or an inconsistently successful transfer where most 

samples transfer successfully but some fail. Red (Figure 32) represents an array with major failures. 

Dark green represents the preferred operating points, light green is generally still acceptable; yellow 

and red are not. 



34 
 

 

Figure 29: Dark Green: A successful transfer with no residual. 

 

Figure 30: Light Green: A successful transfer with some residual. 

 

Figure 31: Yellow: A transfer with high residual. 

 

Figure 32: Red: A failed transfer. 



 

35 
 

This experiment was carried out with several different material combinations, described in Table 2. In 

M2, supplier A electrodes and membrane A were used to find the general trends for the parameter 

space. In M3, M4 and M5 the experiment was repeated with the same electrode materials and 

membranes B, C and D respectively in order to understand the impact of using membranes with 

different ionomers. Note that the ionomer was only varied in the membrane, not the electrode. In M6 

the same membrane and anode as in M2 were used, but with the in-house produced cathode 1. 

Table 2: Description of the experiments for finding the parameter space for the decal transfer process. 

Experiment 
Designation 

Materials Description 

M1 Waste Material 
Test matrix to refine the experiment 

method and setup 

M2 
Membrane A, Supplier A 

anode and cathode 
Finding trends and a baseline 

parameter space 

M3 
Membrane B, Supplier A 

anode and cathode 
Varying the membrane 

M4 
Membrane C, Supplier A 

anode and cathode 
Varying the membrane 

M5 
Membrane D, Supplier A 

anode and cathode 
Varying the membrane (HC PEM) 

M6 
Membrane A, Supplier A 

anode, in-house cathode 1 
Varying the cathode (relative to M2) 

 

Because M2 shows the general trends for the process parameters, some assumptions were made for 

the tests in experiments M3 to M6. First, when a transfer fails (red or yellow), conditions with lower 

temperature or pressure are also unsuccessful. Second, if the operating conditions above and below 

or to both sides of a cell transfer successfully, that cell is also considered successful. These assumptions 

lowered the number of experiments required for a matrix by as much as 2/3. 

 

 

 

 

 

 

 

 

 

 

 

 



36 
 

6.2 Results 
 

The results for the M2 experiment matrix, with supplier A anode and cathode and membrane A, are 

shown in Figure 33. 

 

Figure 33: Process parameter space for M2: Supplier A anode and cathode with membrane A. 

There appears to be a minimum pressing temperature which is one step higher at the lower time level. 

The temperatures in question are 140 and 160 °C, compared to the measured midpoint glass transition 

temperature of 157 °C for this membrane (see chapter 3). At the higher time level, there also appears 

to be a minimum pressure that is required, though the C4 operating point with T=160 °C, p=150 N/cm2 

and t=300 s seems to be an exception. At the lower time level, lower pressures enable a good transfer 

as well if the temperature is sufficiently high. 

The results for the M3 experiment matrix, with supplier A anode and cathode and membrane B, are 

shown in Figure 34. 



 

37 
 

 

Figure 34: Process parameter space for M3: Supplier A anode and cathode with membrane B. 

Once again, the results indicate a minimum temperature requirement which is at 140 to 160 °C. The 

membrane B used here has a glass transition temperature comparable to membrane A at 162 °C. 

Unlike the previous matrix, the pressure boundary is much more similar between the two time levels, 

with the longer time level achieving slightly better transfers at low pressure than the higher time level. 

Another point of interest is that at very high temperatures in the lower time level the transfer worsens 

slightly, although not enough to fully fail. 

The results for the M4 experiment matrix, with supplier A anode and cathode and membrane C, are 

shown in Figure 35.

 

Figure 35: Process parameter space for M4: Supplier A anode and cathode with membrane C. 

The lower bounds for temperature are similar here as for the previous two membranes. The Tg for 

this membrane is not known, as the measurement was unsuccessful (see section 3.1), but is likely 



38 
 

comparable to the Tgs for membranes A and B. At the higher time level, the lower bound for pressure 

is like the previous test with membrane B. At the lower time level, the temperature range for a 

successful transfer is significantly reduced, however, with high temperatures transferring in the yellow 

region. 

The results for the M5 experiment matrix, with supplier A anode and cathode and membrane C, are 

shown in Figure 36. 

 

Figure 36: Process parameter space for M5: Supplier A anode and cathode with membrane D. 

The membrane in this test is a hydrocarbon PEM, with a significantly different structure from PFSA 

PEMs and a measured glass transition at 188 °C. The minimum required temperature is 

correspondingly higher than for the other membranes. The pressure requirement also seems to be 

higher, and at the higher time level seems to be affected by the pressing temperature. At the lower 

time level, the transfer space is significantly smaller than for the higher time level. 

The results for the M6 experiment, with membrane A and supplier A anode material but in-house 

produced cathode 1 material, are shown in Figure 37. 



 

39 
 

 

Figure 37: Process parameter space for M6: Supplier A anode and membrane A with in-house cathode 1. 

The available processing parameters are like what was initially expected, with a minimum temperature 

requirement, which is lower for the lower time level, and a pressure requirement that is lowered with 

increasing temperature in both time levels. 

 

6.3 Discussion 
 

The expected results are as follows: It is expected that there is a minimum required temperature to 

achieve a successful transfer which is related to the glass transition temperature of the ionomer in the 

membrane. It is expected that below a minimum pressure, the compression force on the MEA will be 

too low and the transfer will either fail or be inconsistent. This pressure may be lower at higher 

temperatures as it may relate to the degree of softening of the membrane. The results from section 

4.3 show that at the lower time level (t=120 s) most of the pressing time is spent below the set 

temperature and there should be a correlation between temperature and time whereby a longer time 

may allow for a lower temperature. 

For the lower limit on the temperature, these expectations are met. Each matrix shows a hard limit 

temperature below which the transfer is unsuccessful, with the higher time level having this limit one 

step lower than the lower time. This makes sense, as the section 4.3 results indicate that at the lower 

time level, the MEA only barely reaches the desired temperature. There also seems to be a correlation 

with the glass transition temperature as expected. The test matrices with membrane A (Tg = 157 °C), 

membrane B (Tg = 162 °C) and membrane C – also a PFSA membrane which is expected to have a 

comparable Tg – all have the lowest transfer temperature at 140 °C for the longer time level and 160 

°C for the shorter time. It does not, then, appear necessary to cross the glass transition temperature, 

though it does seem like the transfer temperature must approach the glass transition temperature for 

a successful result. The membrane seems to be softening sufficiently while somewhat below the Tg. 



40 
 

Also interesting is that some of the matrices, particularly M4, show a worse transfer at very high 

temperatures than at more moderate temperatures. This is especially clear at the lower time level. 

This could be a result of thermal ionomer degradation, but in that case the longer time level should 

show this behavior even more strongly than the lower time level. Another reason might be the decal 

substrate softening and having a stronger adhesion to the electrode, though this should also affect 

both temperature levels. The thermal-mechanical behavior of the membrane may warrant further 

investigation. 

In most of the matrices, the pressure performed as expected, with very low pressures rarely 

performing well, and higher pressures performing better, so some pressure seems to be required. The 

limit of the pressure seems to be temperature dependent, though the higher time limit in the M2 

experiment matrix is an exception, with higher temperatures enabling lower transfer pressures. This 

makes sense if the degree to which the ionomer softening increases with temperature. The link 

between the required pressure and the transfer time is considerably opaquer, however. M4 and M6 

show little to no difference in the required pressure between the different time levels; M3 and M5 

suggest that a longer transfer time may allow for lower pressures to function; and M2 suggests that 

having a shorter transfer time enables lower pressures. This last trend is particularly inexplicable. 

These contradictory trends are hard to parse, but overall, it seems that lower pressures are less 

consistent, and that the link between pressure and time is less pronounced compared to the link 

between pressure and temperature or temperature and time. 

Time generally performs as expected. The shorter time level generally indicates a smaller available 

parameter space, and the transfer time is strongly linked to the transfer temperature. A time level 

near or above the glass transition seems to be a significant requirement for a successful transfer, and 

if it becomes necessary to reduce the time it takes to produce an MEA, it may be worthwhile to 

consider times between the values of 120 s and 300 s used here. If a more thermally conductive 

material is used for the pressure distribution pads, this would allow the temperature to more rapidly 

reach the set operating point. Another point of note is that many of the matrices at the lower time 

level had more light green results than the matrices at the longer time level – a longer pressing time 

seems to give more stable and consistent results. 

There is a strong link between the available parameters and the materials used in the transfer. Both 

M2 and M6 use membrane A and supplier A anodes, but for M2 the cathode also comes from supplier 

A while for M6 the cathode was the in-house produced cathode 1. At the lower time level, the matrix 

is nearly identical, and the minimum temperature at the higher and lower time levels respectively is 

also the same. At the higher time level, the higher pressure regions of the graph are also very similar. 

At the lower pressure region, the transfer behavior of the M2 matrix with the supplier A cathode, was 

worse. The in-house cathode achieved better transfer results at low pressure and a long transfer time, 

which is very unexpected. This may be because the M2 matrix is unusual – none of the other matrices 

show a behavior where the low-pressure transfer is worse at the longer transfer time. Another 

possibility is that the in-house cathode had characteristics that favored the transfer at lower pressures, 

such as a rougher dispersion or catalyst particle size distribution. A more likely cause is that the higher 

I/Pt ratio of ionomer to catalyst in the in-house fabricated cathode (I/Pt = 1.0) compared to the 

supplier A cathode (I/Pt = 0.9) enabled these low-pressure transfers. In this case, it may be worth 

investigating in more detail the effects of I/Pt on the available transfer space. 

M2, M3 and M4 all use supplier A anode and cathode, but different PFSA membranes (A, B and C 

respectively). We have already seen how the similar glass transition temperatures of the PFSA 

membranes explain the similar lower temperature bounds. Except for the longer time level of M2, 

they show a similar pressure limit that increases with temperature. Of note is that M3 and particularly 



 

41 
 

M4 show a worse performance at very high temperatures, particularly at the lower time level. This 

behavior does not seem to be related to the side chain length of the membrane’s ionomer or to the 

thickness of the membrane, as in these cases matrix M2 (using membrane A) should show worse 

transfer at very high temperatures than M3 (using membrane B) which is not the case. It is possible 

this may be a result of different thermomechanical behavior or due to differences in the membrane’s 

reinforcement material. Different materials seem to have different requirements for their transfer 

time, which may be a result of them softening to different degrees. 

M5 uses a membrane D, a hydrocarbon PEM with a glass transition temperature at 188 °C. The lower 

bounds for the temperature are accordingly higher than for the test matrices with PFSA membranes, 

at 160 and 180 °C for the higher and lower time levels respectively. The membrane also requires higher 

pressures than the PFSA membranes to transfer well. Such significant differences are expected, as the 

HC ionomer has a fundamentally different structure compared to PFSA ionomers. The higher 

temperature required may make it difficult to pair with PFSA ionomer catalysts if those catalysts have 

low degradation temperatures. 

 

6.4 Conclusion 
 

The materials used in the decal transfer, particularly the membrane, play a significant role in 

determining what operating points are favorable. Of particular importance seem to be the I/Pt ratios 

of the catalyst and the glass transition temperature of the membrane. The glass transition of the 

ionomer in the catalyst layers is also likely to be significant, but is not examined closely here. 

Some general trends can also be established for the influence of the transfer parameters. The 

temperature has a minimum value which must be met for a successful transfer. If the transfer time is 

only barely enough for the temperature to be met, then the required temperature will be higher. Very 

high temperatures can result in worse transfers, however. Higher pressures favor a complete, 

successful transfer, though higher temperatures can allow for lower pressures. This is very dependent 

on the material configuration, and it may be that, within reason, more pressure is better. The transfer 

time must, of course, be sufficient for the actual MEA temperature to reach the necessary levels, 

which is influenced by factors such as the thermal resistance of the pressure distribution pads, but 

more consistent results can be achieved with sufficiently long pressing time. 

Overall, an understanding of the available parameter space within which a successful transfer can be 

consistently achieved has been reached, and some of the design rules for MEA production illuminated. 

This should serve as a good foundation for in-situ performance measurements. 

  



42 
 

7 Performance of the MEAs 
 

It is crucial to understand the effects of the decal transfer processing parameters and the materials 

the MEA is made of on cell performance. The results from chapter 6, which show the available transfer 

space for the different materials, are used to inform the operating points at which MEAs are produced 

for in-situ testing. 

As it is necessary to have a general understanding of the variance for these tests, a qualitative 

statistical evaluation with five repeats is performed. The results from this should give an indication of 

what overall variance can be expected from all sources of error.  

In-situ measurements are also done for the series stacking experiment described in chapter 4.4,