Periodic Reporting for period 2 - ELECTRODE (Exploring the Limits of Mass Transport in Electro-Chemical Energy Converters ThRough uncOnstrained Design and Interface Engineering)
Reporting period: 2022-08-01 to 2024-01-31
Preventing climate change is a major challenge of our society today and electrochemical energy storage devices are key elements for rendering our economy energy efficient. Whereas batteries store energy within their electrodes, fuel cells and flow batteries store it externally in an electrochemical energy carrier (a fuel). Both technologies operate reversibly, i.e. can cycle between electrical and chemical energy, yet only the latter can store energy independent from cell size and proportional to the amount of external energy carrying fluid. Energy conversion then depends on fluid flow processes within the electrodes, which requires transporting and/or mixing fuels, bringing reactant species to the electrolyte, and transporting products away. This challenging task has not been tackled sufficiently so far to bring such green energy concepts to broader application. The core component responsible for transport in all such architectures is the electrode, which must fulfil a multitude of conflicting transport requirements. Electrodes must provide pathways for reactants and products in different states of matter and facilitate dissimilar transport mechanisms. They must be chemically inert, corrosion resistant, provide conflicting surface properties for separating the transport of reactants and products while being catalytically active or able to support anchoring sites for catalytic particles. They must combine high mechanical strength with low weight and volume while being gas tight, crack free, and stable towards oxidation as well as compatible with other components of the assembly. In addition, they must operate reliably over a wide range of temperatures, must be cost effective and easily manufacturable.
The design of an electrode needs to balance convection and diffusion, which are physically fundamentally different processes separated by two orders of magnitude in length scales. The former transport mechanism involves the movement of species by bulk motion and the latter by a gradient in concentration. Convection is related to macroscopic (mm to cm scale) flow (inside the flow channels, FC) and diffusion to the movements of atoms and molecules (µm-nm scale). The complex interaction of multiphase and multiphysics flow in electrode architectures has challenged researchers for half a century.
The state-of-the-art electrode is a multi-layer stack comprising a catalytic layer (CL, site of the chemical reaction), a micro-porous layer (MPL, blocks liquids), a gas diffusion layer (GDL, facilitates diffusion-based transport of reactants and products), flow channels (FC, provides convective fuel flow pathways for reactants), and bipolar plates (BP) that compresses the entire Membrane Electrode Assembly (MEA). However, this architecture is a compromise: to allow for these vastly differing feature ranges, our limited manufacturing abilities require us to confine a set of producible geometries and properties in layers separated by sharp boundaries, whereas the transport mechanisms at play are manifold and spatially co-located not having clear and sharp demarcation lines. For GDL and MPL architectures, we are only able to produce heterogeneous architectures with limited control over their geometries. This entails that researchers first need to discern the random, inhomogeneous, and quite complex internal structure and material properties before studies on mass transport mechanisms can start.
To overcome this impasse, we are changing the starting point of future experiments: Rather than tediously studying mechanisms at play in current designs, we aim to hypothesize and validate our assumptions directly by converting a simulated model into a physical design for subsequent testing. We first model individual transport mechanisms and the correlation between individual structural or chemical changes on the computer, derive 3D architectures on the computer and print them on a 3D printer for subsequent evaluation in a fuel cell testing station. The enabling technology is the fabrication of dimensionally stable carbon from patterned photopolymers. We define dimensionally stable carbon as a type of carbon material that is derived from patterned photopolymers and is converted to carbon in a subsequent heat treatment. During patterning, the photopolymer is cured in a predefined 3D shape through interaction with light of a specific wavelength. Both chemical composition (the formulation) and the heat treatment program (carbonization protocol) is being developed in this project. With this approach, we can derive amorphous carbon electrodes from a Computer Aided Design (CAD) with features spanning over multiple orders of magnitude.
This way, we will omit the conventional necessary step of evaluating single features of electrode architectures and allow researchers to focus on what is relevant: finding the right mechanisms at play and derive corresponding strategies to govern the complex flow characteristics and the electrical, thermal and chemical properties required to construct the next generation of clean energy converters.
Ultimately, the knowledge generated in this project shall render electrochemical flow converters (fuel cells, electrolyzers and flow batteries) more attractive and commonplace in our transition towards clean energy concepts.
The design of an electrode needs to balance convection and diffusion, which are physically fundamentally different processes separated by two orders of magnitude in length scales. The former transport mechanism involves the movement of species by bulk motion and the latter by a gradient in concentration. Convection is related to macroscopic (mm to cm scale) flow (inside the flow channels, FC) and diffusion to the movements of atoms and molecules (µm-nm scale). The complex interaction of multiphase and multiphysics flow in electrode architectures has challenged researchers for half a century.
The state-of-the-art electrode is a multi-layer stack comprising a catalytic layer (CL, site of the chemical reaction), a micro-porous layer (MPL, blocks liquids), a gas diffusion layer (GDL, facilitates diffusion-based transport of reactants and products), flow channels (FC, provides convective fuel flow pathways for reactants), and bipolar plates (BP) that compresses the entire Membrane Electrode Assembly (MEA). However, this architecture is a compromise: to allow for these vastly differing feature ranges, our limited manufacturing abilities require us to confine a set of producible geometries and properties in layers separated by sharp boundaries, whereas the transport mechanisms at play are manifold and spatially co-located not having clear and sharp demarcation lines. For GDL and MPL architectures, we are only able to produce heterogeneous architectures with limited control over their geometries. This entails that researchers first need to discern the random, inhomogeneous, and quite complex internal structure and material properties before studies on mass transport mechanisms can start.
To overcome this impasse, we are changing the starting point of future experiments: Rather than tediously studying mechanisms at play in current designs, we aim to hypothesize and validate our assumptions directly by converting a simulated model into a physical design for subsequent testing. We first model individual transport mechanisms and the correlation between individual structural or chemical changes on the computer, derive 3D architectures on the computer and print them on a 3D printer for subsequent evaluation in a fuel cell testing station. The enabling technology is the fabrication of dimensionally stable carbon from patterned photopolymers. We define dimensionally stable carbon as a type of carbon material that is derived from patterned photopolymers and is converted to carbon in a subsequent heat treatment. During patterning, the photopolymer is cured in a predefined 3D shape through interaction with light of a specific wavelength. Both chemical composition (the formulation) and the heat treatment program (carbonization protocol) is being developed in this project. With this approach, we can derive amorphous carbon electrodes from a Computer Aided Design (CAD) with features spanning over multiple orders of magnitude.
This way, we will omit the conventional necessary step of evaluating single features of electrode architectures and allow researchers to focus on what is relevant: finding the right mechanisms at play and derive corresponding strategies to govern the complex flow characteristics and the electrical, thermal and chemical properties required to construct the next generation of clean energy converters.
Ultimately, the knowledge generated in this project shall render electrochemical flow converters (fuel cells, electrolyzers and flow batteries) more attractive and commonplace in our transition towards clean energy concepts.
In the initial phase of the project, our task was to develop a photopolymer and a corresponding carbonization protocol for dimensionally stable carbon production allowing much larger feature sizes than those reported in literature. For the recreation of features of the flow plates of fuel cells, state-of-the-art dimensions and volumes are insufficient. To comprehend the work performed, an explanation of the process of dimensionally stable carbonization that follow predefined CAD is given.
Dimensionally stable carbon is a type of material required for the rapidly growing field of architected carbon, where extreme properties are derived from geometry and material properties concurrently. To obtain this material, particular photopolymerizable formulations are cured to a predefined shape by means of a light pattern. Then, the pre-shaped photopolymeric part undergoes a heat treatment, in which the polymer thermally degrades, and converts to amorphous carbon shrinking equally in all directions thereby undergoing a weight and volume reduction. The relative dimensions are kept.
Photopolymers inevitably contain species that cannot be converted to carbon. These “non-carbonizable species” are removed from the polymer network during heat treatment. They exit the polymer network during its conversion to carbon through diffusion. This becomes increasingly difficult as a larger volume is carbonized as the “non-carbonizable” species need to diffuse longer distances to exit the polymer network. A larger volume has larger mass; more weight is carried by the degrading polymer rendering a higher stiffness of the degrading polymer necessary. This in turn compromises diffusivity, i.e. the polymer’s ability to transport “non-carbonizable” species. The larger the volume (more specifically: the lower the surface to volume ratio) of a part, the more demanding it is to retain the dimensional stability during carbonization.
In the initial phase of the project, we screened several photopolymers and carbonization protocols for their suitability for large volume carbonization. We found that heavily crosslinked, thermally stable photopolymer with large carbon and large aromatic content are suitable candidates for a successful dimensionally stable precursor. We also developed a carbonization protocol and corresponding theory allowing us the production of parts with several mm^2 of volume and three orders of magnitude lower surface to volume ratios than those reported in literature. We obtained features with dimensions differing by over two orders of magnitude, which are routinely converted to dimensionally stable carbon without changing geometric relations. For the features producible in this project, this is sufficient. However, we discovered another formulation and developed a dedicated carbonization protocol providing even better outcomes. Based on the current understanding, this formulation should not yield a high carbon yield (= amount of carbon atoms in the resulting amorphous carbon part divided by the amount of carbon atoms in the polymer network before the thermal treatment). With this formulation, we can produce dimensionally stable carbon of several cm in size, while still maintaining a resolution below 100 µm. This approach could be a game changer in the field, and we are currently preparing a respective publication.
We worked on theoretically determining porosity-dependent permeability, electrical and thermal conductivity of so called triply periodic minimal surfaces (TPMS; primitive, diamond, gyroid…) and compared these properties to those of a simple cubic lattice structure. We assessed the structure that best performs as a defined and homogeneous porous network in a hypothetical gas diffusion layer with homogeneously distributed unit cells. A Schwartz P type lattice and the simple cubic lattice showed up to one order of magnitude higher permeabilities than walled structures such as Schoen Gyroid and Schwatz P structure. However, the walled structures have on average 1.3 and 2.6 times the electrical and thermal conductivity of their lattice type counterparts, respectively. The properties of the lattice structures were largely affected by the surface area density (amount of surface area in a given volume), whereas tortuosity (structural deviations from straight lines) variation did not impact permeabilities and conductivities to a large extent. An analytical relation was set up to allow future designers of such porous media to fine-tune porosity according to electrical and thermal conductivity requirements. As permeability is the key element to tune regarding liquid water transport in gas diffusion layers, it was decided that simple cubic lattices will be produced initially. These structures are the simplest to fabricate with our 3D printer at high resolution.
We developed a 2D framework allowing for benchmarking the effect of simple design changes (lattice spacing, void volume, layer spacing, strut dimensions, etc.) in a two-phase flow scenario. A convective flow across the top of a homogenous lattice structure resembles a flow field on top of a gas diffusion layer and water entering the lattice from underneath resembles the water production on top of the electrolyte membrane of a polymer electrolyte fuel cell. The computational time of this framework is acceptable, and results are meaningful to capture the effect of the lattice geometry on the water and gas transport. The model allows to validate hypothesis relatively quickly capturing the transport effects of the most important design criteria. A publication in combination with experimental results on the fuel cell tests is under way.
To save computational time, a Lattice Boltzmann Method based simulation tool is developed in which mathematically defined porous media can be imported without meshing (the approximation of smooth geometry in regular finite elements). This saves computation time and tedious preprocessing in comparison to conventional CFD modelling. Also, problems with higher complexity may be modelled in an acceptable duration. We see great potential in this tool for large multiscale and Multiphysics simulations; however, we could not yet validate it and are trying to resolve remaining issues.
We performed current-voltage tests with conventional Polymer Electrolyte Membrane Fuel Cells and compared them to equivalent cells, comprising 3D printed gas diffusion layers and to cells comprising combined 3D printed flow fields and gas diffusion layers. The ohmic resistance of cells including our 3D printed parts seems in range with reference cells. In the low current density regime, the performance is state of the art. The active area utilization (the degree to which the area of the cell covered with catalyst) was poor in our first attempts as some features of our printed features shielded the area. This was resolved. In the higher current density regimes, our electrode parts are lower performing. We believe this to be a resolvable technical limitation, which we will detail in an upcoming article.
Dimensionally stable carbon is a type of material required for the rapidly growing field of architected carbon, where extreme properties are derived from geometry and material properties concurrently. To obtain this material, particular photopolymerizable formulations are cured to a predefined shape by means of a light pattern. Then, the pre-shaped photopolymeric part undergoes a heat treatment, in which the polymer thermally degrades, and converts to amorphous carbon shrinking equally in all directions thereby undergoing a weight and volume reduction. The relative dimensions are kept.
Photopolymers inevitably contain species that cannot be converted to carbon. These “non-carbonizable species” are removed from the polymer network during heat treatment. They exit the polymer network during its conversion to carbon through diffusion. This becomes increasingly difficult as a larger volume is carbonized as the “non-carbonizable” species need to diffuse longer distances to exit the polymer network. A larger volume has larger mass; more weight is carried by the degrading polymer rendering a higher stiffness of the degrading polymer necessary. This in turn compromises diffusivity, i.e. the polymer’s ability to transport “non-carbonizable” species. The larger the volume (more specifically: the lower the surface to volume ratio) of a part, the more demanding it is to retain the dimensional stability during carbonization.
In the initial phase of the project, we screened several photopolymers and carbonization protocols for their suitability for large volume carbonization. We found that heavily crosslinked, thermally stable photopolymer with large carbon and large aromatic content are suitable candidates for a successful dimensionally stable precursor. We also developed a carbonization protocol and corresponding theory allowing us the production of parts with several mm^2 of volume and three orders of magnitude lower surface to volume ratios than those reported in literature. We obtained features with dimensions differing by over two orders of magnitude, which are routinely converted to dimensionally stable carbon without changing geometric relations. For the features producible in this project, this is sufficient. However, we discovered another formulation and developed a dedicated carbonization protocol providing even better outcomes. Based on the current understanding, this formulation should not yield a high carbon yield (= amount of carbon atoms in the resulting amorphous carbon part divided by the amount of carbon atoms in the polymer network before the thermal treatment). With this formulation, we can produce dimensionally stable carbon of several cm in size, while still maintaining a resolution below 100 µm. This approach could be a game changer in the field, and we are currently preparing a respective publication.
We worked on theoretically determining porosity-dependent permeability, electrical and thermal conductivity of so called triply periodic minimal surfaces (TPMS; primitive, diamond, gyroid…) and compared these properties to those of a simple cubic lattice structure. We assessed the structure that best performs as a defined and homogeneous porous network in a hypothetical gas diffusion layer with homogeneously distributed unit cells. A Schwartz P type lattice and the simple cubic lattice showed up to one order of magnitude higher permeabilities than walled structures such as Schoen Gyroid and Schwatz P structure. However, the walled structures have on average 1.3 and 2.6 times the electrical and thermal conductivity of their lattice type counterparts, respectively. The properties of the lattice structures were largely affected by the surface area density (amount of surface area in a given volume), whereas tortuosity (structural deviations from straight lines) variation did not impact permeabilities and conductivities to a large extent. An analytical relation was set up to allow future designers of such porous media to fine-tune porosity according to electrical and thermal conductivity requirements. As permeability is the key element to tune regarding liquid water transport in gas diffusion layers, it was decided that simple cubic lattices will be produced initially. These structures are the simplest to fabricate with our 3D printer at high resolution.
We developed a 2D framework allowing for benchmarking the effect of simple design changes (lattice spacing, void volume, layer spacing, strut dimensions, etc.) in a two-phase flow scenario. A convective flow across the top of a homogenous lattice structure resembles a flow field on top of a gas diffusion layer and water entering the lattice from underneath resembles the water production on top of the electrolyte membrane of a polymer electrolyte fuel cell. The computational time of this framework is acceptable, and results are meaningful to capture the effect of the lattice geometry on the water and gas transport. The model allows to validate hypothesis relatively quickly capturing the transport effects of the most important design criteria. A publication in combination with experimental results on the fuel cell tests is under way.
To save computational time, a Lattice Boltzmann Method based simulation tool is developed in which mathematically defined porous media can be imported without meshing (the approximation of smooth geometry in regular finite elements). This saves computation time and tedious preprocessing in comparison to conventional CFD modelling. Also, problems with higher complexity may be modelled in an acceptable duration. We see great potential in this tool for large multiscale and Multiphysics simulations; however, we could not yet validate it and are trying to resolve remaining issues.
We performed current-voltage tests with conventional Polymer Electrolyte Membrane Fuel Cells and compared them to equivalent cells, comprising 3D printed gas diffusion layers and to cells comprising combined 3D printed flow fields and gas diffusion layers. The ohmic resistance of cells including our 3D printed parts seems in range with reference cells. In the low current density regime, the performance is state of the art. The active area utilization (the degree to which the area of the cell covered with catalyst) was poor in our first attempts as some features of our printed features shielded the area. This was resolved. In the higher current density regimes, our electrode parts are lower performing. We believe this to be a resolvable technical limitation, which we will detail in an upcoming article.
The discoveries made in the field of dimensionally stable carbon in this project can be considered disruptive. The current dimensions reach beyond our expectations and beyond the current state-of-the-art. The achievable features are larger than would be required for any part of the project moving forward. Yet, we see great value in developing this technology further, independent of the field of application. Certain obtained results contradict statements in literature, and we are currently investigating the mechanisms behind leading to this achievement.
At the end of the project, we believe to have built a prototyping platform able to print any 3D part comprising sub 100 µm scale features and volumes in the cubic cm regime. The parts will be dense and defect free. We aim to detail the full thermal treatment and chemistry necessary for achieving dimensionally stable carbon at the cubic cm scale. Further we aim to explain the underlying mechanisms during carbonization and detail the structure-property relationships to tune dimensional stability, mechanical, chemical, and electrical properties.
The work on computer fluid dynamics modelling will soon be submitted for publication in conjunction with experimental fuel cell performance tests. The framework we are currently working on is a solid basis for hypotheses evaluation, however, does not exceed the state-of-the-art methodically. The developed LBM tool is innovative; it is however not validated yet. By the end of the project, we can utilize this method for simple problems and compare its performance to conventional CFD tools. A follow-up project involving CFD experts on LBM is planned.
As we have a significant range of feature sizes at our disposal, we started working on electrode architectures comprising features of several layers consolidated in one part. This is beyond the state-of-the-art in the field of carbon electrodes. We anticipate the disruptively redesign of the entire electrode architecture carefully reconsidering all current features. We aim to provide and take on a range of hypotheses on transport mechanisms and corresponding architectures that could thus far not been validated and validate them experimentally.
The work on replication of MPL and CL with our approach as well as the coating and modifying of our carbon parts for catalytic activity will intensify in the second part of the project. We will focus mainly on Pt coating leveraging from promising results obtained in another project.
At the end of the project, we believe to have built a prototyping platform able to print any 3D part comprising sub 100 µm scale features and volumes in the cubic cm regime. The parts will be dense and defect free. We aim to detail the full thermal treatment and chemistry necessary for achieving dimensionally stable carbon at the cubic cm scale. Further we aim to explain the underlying mechanisms during carbonization and detail the structure-property relationships to tune dimensional stability, mechanical, chemical, and electrical properties.
The work on computer fluid dynamics modelling will soon be submitted for publication in conjunction with experimental fuel cell performance tests. The framework we are currently working on is a solid basis for hypotheses evaluation, however, does not exceed the state-of-the-art methodically. The developed LBM tool is innovative; it is however not validated yet. By the end of the project, we can utilize this method for simple problems and compare its performance to conventional CFD tools. A follow-up project involving CFD experts on LBM is planned.
As we have a significant range of feature sizes at our disposal, we started working on electrode architectures comprising features of several layers consolidated in one part. This is beyond the state-of-the-art in the field of carbon electrodes. We anticipate the disruptively redesign of the entire electrode architecture carefully reconsidering all current features. We aim to provide and take on a range of hypotheses on transport mechanisms and corresponding architectures that could thus far not been validated and validate them experimentally.
The work on replication of MPL and CL with our approach as well as the coating and modifying of our carbon parts for catalytic activity will intensify in the second part of the project. We will focus mainly on Pt coating leveraging from promising results obtained in another project.