WO2020173823A1

WO2020173823A1 - Non-invasive method and system for local ionomer impedance determination in polymer electrolyte fuel cells

Info

Publication number: WO2020173823A1
Application number: PCT/EP2020/054588
Authority: WO
Inventors: Jens ELLER; Arnaud SCHULLER; Valentina STAMPI-BOMBELLI
Original assignee: Paul Scherrer Institut
Priority date: 2019-02-27
Filing date: 2020-02-21
Publication date: 2020-09-03
Also published as: EP3931898A1

Abstract

The present invention discloses a method and a system for the determination of the impedance of a polymer electrolyte fuel cell (8), said method comprising the steps of: a) attaching a plurality of electrical contacts (En) to one or a pair of conductive layers (2, 4) between which the ionomer layer (6) is sandwiched; b) applying an AC current over pairs of the electrical contacts (En) of said plurality of electrical contacts (En) according to a determined AC current application scheme; c) measuring the voltage response to the applied AC current over pairs of electrical contacts (En) of said plurality of electrical contacts (En) according to a determined voltage measuring scheme; and d) evaluating the measured voltage response according to the determined voltage measuring scheme in dependency of the determined application scheme; and e) determining from the evaluated measure voltage responses the impedance of the electrochemical cell, preferably as a conductivity distribution over the ionomer layer (6); and f) preferably comparing the determined impedance values in dependency of the determined application scheme with a number of known impedance patterns.

Description

Non-invasive method and system for local ionomer impedance determination in polymer electrolyte fuel cells

The present invention relates to a non-invasive method and system for ionomer conductivity determination in polymer electrolyte fuel cells.

Polymer Electrolyte Fuel Cells (PEFC) are an essential

technology for future clean, fast refill, and long-range electric mobility. For commercial success however, PEFC technology still needs improvements in performance, durability and cost. Increasing specific power will reduce cost, however, at high current density operation, the water management limits efficiency and performance. In the channels of flow field (FF) plates of technical sized cells with active areas of some hundred cm², there is an accumulation of product water from the gas inlet towards the gas outlet.

These large-scale gradients cause changes in the ionic

conductivity of the polymer electrolyte membrane of the cell and within the ionomer in the catalyst layer, which then give rise to current density variations, leading to lower cell efficiency and shorter lifetime. Therefore, the understanding and measurement of operando membrane conductivity in large technical cells is of great importance. Indeed, it would provide information both on the degradation state as well as the performance.

Until now, the determination of the cells in-plane ionomer conductivity distribution has almost exclusively been solved in an invasive way making use of either segmented flow field plates or invasive boards that are placed between individual cells. One such technique uses a shunt board inserted between the cell's flow field plate and current collector plate or between the cells of a fuel cell stack. This method is widely used for current distribution measurement and does also account for the ionomer conductivity in the membrane. Another related method exploits the use of segmented flow fields to apply locally resolved electrochemical impedance spectroscopy. This set of techniques provides useful information regarding ionomer conductivity and current distribution. However, they are invasive and therefore not suited for actual applications.

The determination of the current density distribution has been also achieved by magnetic field sensors. Since the current distribution of the cell or stack induces an electromagnetic field, it is also possible to measure it at different

locations and to reconstruct the actual current distribution by solving an inverse problem. This method has been applied both in an invasive and non-invasive way. While this method showed promising results in differentiating healthy and faulty cells in a stack, and detecting different current density scenarios, it does not provide information on the ionomer conductivity .

There exist also other conductivity measurement methods for fuel cells that provide only conductivity information averaged over the whole cell such as electrical impedance spectroscopy, high frequency resistance measurement or the current

interruption/pulsing method.

It is therefore the objective of the present invention to provide a non-invasive method which is able to provide

localized information about the ionic conductivity in the membrane and catalyst layer that could be used both in

research as well as in actual application such as the

detection of a degraded cell area within the fuel cell stack in a fuel cell electric vehicle.

This objective is achieved according to the present invention with respect to the method by the feature given in claim 1. With respect to the system, this objective is achieved by a system according to the features of claim 7.

Preferred embodiments of the present invention are listed in claims 2 to 6 and 8 to 12.

The present invention therefore uses stimulation AC currents, preferably at different frequencies, to implement a non- invasive localized electrochemical impedance spectroscopy. The method and the system advantageously allow for spatially resolved differentiation of at least one of membrane and catalyst layer ionomer conductivity, current density

distribution, water flooding detection, fuel starvation, humidity distribution, catalyst layer degradation and state of health of the membrane. The measurement and evaluation process can be implemented such that the conductivity measurement gives real time responses.

Typically, the electrochemical cell comprises a polymer electrolyte membrane and a catalyst layer in contact with at least one side of the polymer electrolyte membrane.

In a preferred embodiment of the present invention, the AC current application scheme may comprise the application of AC currents to a number of changing pairs of electrical contacts and/or the application of AC currents having different current levels and/or different frequencies. The AC current

application scheme can be implemented by the application of a multiplexer which connects an AC current source subsequently to different pairs of the electrical contacts. The pairs of electrical contacts can be for example adjacent contacts and/or two contacts being disposed on the same side of the flow field plates and/or two contacts being disposed on opposite sides or other sides of the flow field plates. The exact flow of the AC current application scheme also depends on the desired precision of the evaluation for the impedance, i.e. conductivity, and/or the other determination parameters.

Further, it is advantageously when the voltage measuring scheme is aligned with the AC current application scheme and comprises the measurement of the voltage responses over different pairs of electrical contacts. For example, the voltage measuring scheme may comprise the measurement of the voltage response including amplitude, frequency and phase shift compared to the AC signal for different pairs of electrical contacts while the AC current is applied to one or multiple pair(s) of electrical contact (s) . For the measurement of the voltage responses, the pairs of electrical contacts can be for example adjacent contacts and/or two contacts being disposed on the same side of the flow field plate and/or two contacts being disposed on opposite sides or other sides of the flow field plates. The exact flow of the voltage measuring scheme also depends on the desired precision of the evaluation for the impedance, i.e. conductivity, and/or the other

determination parameters.

In order to speed up the evaluation of the measured voltage responses, the evaluation of the measured voltage responses may comprise a comparison of the measured voltage responses with a number of known voltage response patterns. These voltage response patterns each are assigned to one or more tomographic conductivity distributions over the ionomer layer or to one or more humidity distributions over the cell.

Preferred embodiments of the present invention are hereinafter described in more detail with reference to the attached drawings which depict in:

Fig. 1 schematically the method for the determination of the conductivity of an ionomer layer of a polymer electrolyte fuel cell;

Fig. 2 schematically a simplified FEM model of a fuel cell comprising an ionomer layer with a plurality of electrodes ;

Fig. 3 in a simplified manner the schematic of the EIT

hardware (EIT = Electrical Impedance Tomography) ; and

Fig. 4 schematically the results of a numerical feasibility study comparing a predefined conductivity distribution, an initial guess and a final solution of an EIT solver.

Fig. 5 a) example of 6 measurement patterns done at 3

different cell conditions: 50 % RH at inlet on both anode and cathode (RH A50/C50), 90% RH at inlet on both anode and cathode (RH A90/C90), counter-flow of 50% RH on the anode side and 90% RH on the cathode side (RH A50/C90); and b) results of the interpolation between 50% RH on both sides and 90%

RH on both sides to find out the RH profile in the 50% anode and 90% cathode counterflow case.

The proposed method to reach the objective of a non-invasive method for electrochemical cell, and in particular ionomer, impedance determination, i.e. conductivity determination, in particular in polymer electrolyte fuel cells, is based on electrical impedance tomography (EIT) . Of specific interest is an ionomer layer of the fuel cell which usually comprises a polymer electrolyte membrane being sandwiched between catalyst layers. EIT is a non-invasive method that allows the

determination of the conductivity in the ionomer layer by using electrodes attached to the surface of the flow field plates and solving the inverse problem of the relationship between the flow field plates' surface potential distribution and the impedance distribution in the cell, respectively the conductivity distribution in the ionomer layer.

The flow field plates' surface potential distribution is measured by the injection of an alternating current (with a frequency of a few to 100 kHz) in the low mA range between two or more electrodes and measurement of the potential

distribution at at least some of the remaining electrodes (see Fig. 1) together with a systematic variation of the injection and measuring electrodes. In the example of Fig. 1, fourteen electrodes Ei to E₁₄ are attach to two flow field plates 2, 4 which sandwich the ionomer layer 6 (see Figure 2) . An AC current i_n is supplied to a pair of electrodes Ei and E₂. The electrode Ei is attached to the flow field plate 2; E₂ can be attached to the same flow field plate 2 but could be also attached to the opposite flow field plate 4 (the same applies to the other electrodes E₃ to E ₄) .

Diverse voltage measurements ui to U₉ are performed in response to the application of the current i_n. These voltage

measurements can be repeated for different AC currents i_n+i, i_n+2 and so on wherein the different AC currents are

distinguished by a different frequency. Of course, the AC current application scheme can also provide for the

application of a number of AC currents to another pair of the electrodes while the voltage response measurement scheme then provides for the measurement of the voltage response over at least some of the respective remaining pairs of electrodes. In general, both the AC current application scheme and the voltage response measurement scheme can have a huge variety in terms of AC current application and voltage response

measurement cycles which depend largely from the dimension of the ionomer layer 6 and/or the number of electrodes E_n attached to the flow field plates 2, 4 and/or the position of the attached electrodes E_n and/or the desired accuracy and so on.

Using the different stimulation frequencies non-invasive localized electrochemical impedance spectroscopy is

implemented. It allows for spatially resolved differentiation of membrane and catalyst layer ionomer conductivity, current density distribution, water flooding detection, fuel

starvation and/or catalyst layer degradation. The measurement and evaluation process can be implemented such that the conductivity measurement gives real time responses.

The present invention comprises both a software aspect and a hardware aspect. The software aspect is based on a geometric model (mesh) of a PEM fuel cell as exemplarily shown in Fig.

2. This model is then exploited by the finite element method; while some other methods such as the boundary element method or the mesh free Galerkin method can also be used. In this example, a plurality of 28 electrodes are attached to the flow field plates 2; 14 electrodes to the upper flow field plate 2 and 14 electrodes to the bottom flow field plate 4.

This model includes the ionomer layer 6 of the fuel cell as well as the electrodes E_n attached to the flow field plates 2,

4 (additional details such as gas diffusion layers, micro- porous layers, flowfield and cooling fluid channels, gaskets, subgaskets and fluid manifolds can be considered as well) . The equations governing EIT and correlating boundary surface voltage with the inner conductivity of the membrane and other layers are applied to the model. This allows solving the so- called forward problem and calculating the boundary voltage for any membrane conductivity.

This data is then used to solve the so-called inverse problem to retrieve the fuel cell's local impedance from a given voltage surface data. In EIT applications, inverse problems are usually ill posed. However, this limitation can be tackled by incorporating prior knowledge about cells geometry and the conductivity of ionomer free fuel cell components (flow field plates, Gas Diffusion Layers, micro-porous layers, sub

gaskets, gaskets) into the model.

Therefore, the EIT inverse problem becomes simpler because the area where the conductivity has to be retrieved is much smaller compared to the size of the cell and the conductivity inside that area has known predefined upper and lower limits.

Two specific methods are exemplarily considered to reconstruct the conductivity data from the boundary voltage information u± :

Optimization approach: this method minimizes the

difference between the boundary voltage measured on the actual cell with a forward solver based distribution of boundary voltage values. The difference between the measurement and the forward solver solution is then minimized in an iterative process probing different impedance distributions. Different minimization type solvers can be used, where evolutionary strategy based solvers (e.g. CMAES) have shown to solve the problem in reasonable time.

Neural network approach: this method consists in creating a large dataset with many possible impedance scenarios, preferably for one or more generic types of ionomer layer, and their associated boundary voltages. It is then used as a training set for the neural network that will learn the correlation between boundary voltage and inner conductivity. After initial training, this method allows for a much faster solution, probably at the expense of a small loss in accuracy.

Finally, both methods can be combined using the neural network solution as an initial guess in the optimization algorithm, thereby reducing the computational time of the optimization algorithm.

Since a fine discretization can result in a large number of degrees of freedom to solve for, the problem can be simplified by using ID (e.g. between cell inlet to outlet) or 2D (e.g. for the cells in-plane orientation) parameterizations of the cells impedance distribution which describe the conductivity distribution by fewer degrees of freedom. Possible

parameterizations could be discrete or smooth interpolations between selected points in the membrane domain of the cell.

The hardware aspect for the determination of the conductivity of the ionomer layer 6 in a fuel cell 8 is shown as a system S in Figure 3. The system S includes the electrodes E, a current source CS to supply an AC current signal to pairs of the electrodes E and voltmeters VM (frequency response analyzer (FRA)) to measure the voltage response over pairs of the electrodes E as shown in Fig. 3. Current source and FRA might be implemented within one evaluation and determination unit EDU, have multi-channel options to connect to many electrodes E or use a multiplexer unit for connecting to the different electrodes E. The number of electrodes E can vary between four and a few hundred depending on the available surface of the cell, the desired spatial resolution and the desired accuracy. The number of stimulation pattern (pair of injection

electrodes associated to a pair of measurement electrodes) depends on the number of electrodes, the desired spatial resolution and the desired accuracy and is therefore between one and a few thousand.

The AC currents i_n are applied according to an AC current application scheme that can be administrated in the evaluation and determination unit EDU. The AC current application scheme exemplarily is a table of an AC current application work flow which defines the order of pairs of electrodes E_n and the respective AC currents applied to theses pairs. The voltage response measurement scheme therefore can also be

administrated in the evaluation and determination unit EDU and defines the voltage measuring work flow in response to the application of the AC currents i_n according to the AC current application scheme.

Figure 3 schematically shows a simplified example with six electrodes Ei to Ee. The AC current application scheme can start with the application of an AC current ii = 10 mA * e^iGJt having a frequency of 1kHz which is applied over the

electrodes Ei and E4. The voltage response u(t) can be measured over various pairs of electrodes. Further, the AC current can be varied to have a frequency of 0. lHz to 50kHz; the voltage response is measured accordingly. The same AC current can then be applied to another pair of electrodes and the voltage responses can be measured at various pairs again.

All measurements deliver a landscape of conductivity values that can be compared to a library of learned landscapes having a known conductivity distribution. This comparison enables the evaluation and determination unit to determine the best match of the actually measured conductivity landscape to one

landscape in the library. As a result, a distinct

determination on the conductivity distribution over the ionomer layer 6 can be made by this non-invasive process. Of course, the accuracy of the results depends on a number of parameters, such as the number of electrodes, the position of the electrodes, the AC current application scheme, the voltage response measuring scheme and the number of known conductivity landscapes learned by the library.

Presently, the algorithm based approach has been developed to automatically select the most relevant stimulation patterns for a given cell geometry and electrode configuration to achieve sufficient solution quality in all domains of the electrochemical cell 8. Some more general and well established stimulation patterns can also be used. These include the adjacent pattern, the opposite pattern and the zig-zag

pattern. In those three cases, there is one layer of N

electrodes on each flow field plate. In the adjacent

stimulation patter, the current is injected between a pair of adjacent electrodes (El to E2 in Figure 2 for example) and the resulting voltage is measured between the remaining 2N-3 adjacent pairs (E3 to E4, E4 to E5 etc...).

This stimulation pattern corresponds mainly to in-plane measurements. In the opposite stimulation pattern, the current is injected between a pair of neighbor electrodes on opposite flow field plates (El to E15 for example) and the resulting voltage is measured between the remaining N-l neighboring opposite pairs (E2 to E16, E3 to E17 etc...) . This method corresponds to through plane measurements. Finally, in the zigzag stimulation pattern, the current is injected between electrodes on opposite flow fields (for example first El to E15 and then E15 to E2) . The resulting voltages are measured between the remaining electrodes according to the same scheme (E2 to El 6, El 6 to E3 etc...) .

The following examples in Figure 4 show the reconstruction of a simulated conductivity profile on a single cell with 32 electrodes attached to its surface: six along its length and two along its width on both sides and both flow field plates similar as shown in Figure 2. For cases focusing on simple ID or 2D conductivity distribution scenarios between cell inlet and outlet, a very good agreement between the predefined and EIT determined conductivity distributions has been achieved in a numerical feasibility test scenario (see Fig. 4a-e for ID distributions; Fig. 4f for 2D distributions) where the

measured and evaluated tomographic conductivity distribution almost exactly matches the simulated one.

Another method to reconstruct the conductivity profile does not rely on the FEM model. A stimulation pattern is applied to the different electrodes and the voltage response is measured for many different conditions. This will create a library of known voltage responses corresponding to known humidity profiles. When a new voltage response set is measured, a simple interpolation between known profiles will lead to the corresponding humidity profile. This method allows for a reasonably accurate determination when the precise knowledge of the fuel cell geometry and the different layers

conductivities cannot be obtained.

Figure 5 a) shows an example of multiple measurements done under different humidity conditions, symmetric 50% RH at anode and cathode inlet, symmetric 90% RH at anode and cathode inlet and counter flow 50% RH on one inlet and 90% RH on the other inlet. The stimulation patterns are chosen to represent the impedance at different locations along the cell's channels. Stimulation pattern 1 to 6 represent the impedance from the cell inlet to the cell outlet respectively. It is then

possible to use the resistance measurements done under

symmetric 50% and 90% RH to interpolate the results obtained under mixed 50% / 90% RH conditions in order to determine the humidity profile as shown in figure 5 b) .

This method can be applied using much more stimulation

patterns and more complicated interpolation scheme (than just linear) in order to determine the humidity profile more accurately. Using known humidity dependent parameterizations of the membrane conductivity the membrane conductivity profile can be derived from the humidity profile.

Claims

Patent Claims

1. A non-invasive method for a determination of the local impedance of an ionomer layer (6) in a polymer electrolyte electrochemical cell, such as a fuel cell (8), said method comprising the steps of:

a) attaching a plurality of electrical contacts (E_n) to one or a pair of conductive layers (2, 4) between which the ionomer layer (6) is sandwiched;

b) applying AC currents (i_n) over pairs of the electrical contacts (E_n) of said plurality of electrical contacts (E_n) according to a determined AC current application scheme;

c) measuring the voltage response (u_n) to the applied AC current (i_n) over pairs of electrical contacts (E_n) of said plurality of electrical contacts (E_n) according to a determined voltage measuring scheme; and

d) evaluating the measured voltage response (u_n) according to the determined voltage measuring scheme in dependency of the determined current application scheme; and

e) determining from the evaluated voltage responses the local impedance of the electrochemical cell, preferably as a

tomographic conductivity distribution over the ionomer layer (6) .

2. The method according to claim 1 wherein the local impedance information is interpreted to gain insight on different fuel cell properties, such as catalyst layer ionomer conductivity, water flooding detection, fuel starvation, humidity

distribution, catalyst layer degradation and state of health of the membrane.

3. The method according to claim 1 or 2, wherein

the electrochemical cell comprises a polymer electrolyte membrane and a catalyst layer in contact with at least one side of the polymer electrolyte membrane.

4. The method according to any of the preceding claims, wherein the AC current application scheme comprises the application of the AC currents (i_n) to a number of changing pairs of

electrical contacts (E_n) and/or the application of AC currents having different current levels and/or different frequencies and/or superposing currents with different frequencies.

5. The method according to any of the preceding claims, wherein the voltage measuring scheme is aligned with the AC current application scheme and comprises the measurement of the voltages (u_n) over different pairs of electrical contacts (E_n) .

6. The method according to any of the preceding claims, wherein the evaluation of the measured voltage responses (u_n) comprises a comparison of the measured voltage responses (u_n) with a number of known voltage response patterns.

7. A system (S) for a non-invasive determination of the impedance of a polymer electrolyte electrochemical cell, such as a fuel cell (8), comprising:

a) a plurality of electrical contacts (E_n) attached to one or a pair of conductive layers (2, 4) between which the ionomer layer (6) is sandwiched;

b) an AC current driving circuitry (CS) being enabled to apply AC currents (i_n) over pairs of the electrical contacts (E_n) of said plurality of electrical contacts (E_n) according to a determined AC current application scheme;

c) a voltage measuring circuitry (VM) being enabled to measure the voltage response (u_n) to the applied AC currents (i_n) over pairs of electrical contacts (E_n) of said plurality of

electrical contacts (E_n) according to a determined voltage measuring scheme; and

d) an evaluating unit (EDU) being enabled to evaluate the measured voltage response (u_n) according to the determined voltage measuring scheme in dependency of the determined application scheme; and

e) a determining unit (EDU) being enabled to determine from the evaluated measure voltage responses the local impedance of the ionomer layer (6), preferably as a tomographic

conductivity distribution over the ionomer layer (6) .

8. The system according to claim 7, wherein the local

impedance information is interpreted to gain insight on different fuel cell properties, such as catalyst layer ionomer conductivity, water flooding detection, fuel starvation, humidity distribution, catalyst layer degradation and state of health of the membrane.

9. The system (S) according to claim 7 or 8, wherein

the ionomer layer (6) comprises a polymer electrolyte membrane and a catalyst layer in contact with at least one side of the polymer electrolyte membrane.

10. The system (S) according to any of the preceding claims 7 to 9, wherein

the AC current application scheme comprises the application of the AC currents (i_n) to a number of changing pairs of

electrical contacts (E_n) and/or the application of AC currents (i_n) having different current levels and/or different

frequencies .

11. The system (S) according to any of the preceding claims 7 to 10, wherein the voltage measuring scheme is aligned with the AC current application scheme and comprises the

measurement of the voltages (u_n) over different pairs of electrical contact (E_n) .

12. The system (S) according to any of the preceding claims 7 to 11, wherein the evaluation of the measured voltage

responses (u_n) comprises a comparison of the measure voltage responses (u_n) with a number of known voltage response

patterns .