A Segmented Cell Measuring Technique for Current Distribution Measurements in Batteries, Exemplified by the Operando Investigation of a Zn-Air Battery

Silver (Ag) electrodes consisting of 90 wt.% Ag (Ferro AG, type 311) and 10 wt.% PTFE (TF9207Z, 3 M Dyneon) were produced. For this purpose, the powders were physically mixed in a knife-mill, filled into a mossy frame and pressed at 300 kN m⁻² for 1 min in a hydraulic press together with a metal mesh (0.125 mm, Haver & Boecker). After removal of the frame the electrode was pressed again using the same parameters. To achieve a mechanical stable electrode, it was heat treated in an oven for 1 h at 340 °C as described by Wittmaier et al. ³³ A punching tool with a diameter of 12 mm was used to obtain the electrodes for cell assembly. Ag was used as catalyst due to its high activity and thus, to avoid effects ascribed to inhomogeneous electrode performance.

80 wt.% Timcal Timrex T150 with 20 wt.% PTFE (TF9207Z, 3 M Dyneon) were physically mixed using a knife mill for 30 s. The powder mixture was filled into a frame, calendared at 30 Kg and heat treated for 1 h at 340 °C.

The test cell was build using Poly(methyl methacrylate) (PMMA). It consists of a solid back plate on the anode side, a ring-shaped electrolyte compartment with a thickness of 1 cm and a plate with four connections for gas supply. The gas supply is placed at the four electrode corners (see Fig. 1).

Zoom In Zoom Out Reset image size

A zinc foil (250 μm thick) was used as anode. The produced Ag electrodes were pressed into a PTFE frame and mounted into the cell, together with the GDL and the segmented current collector. As segmented current collector a PEEK foil (100 μm thickness) with printed gold contacts (40 μm thickness) was used. Airflow was realized through the segmented current collector by four holes marked in Fig. 1c as red circles.

The cathode consisted of the 12 mm Ag electrodes pressed inside a 0.5 mm thick PTFE frame. Between the electrolyte and the cathode assembly a rubber foil and a PMMA plate, both with 11 mm holes were mounted. This was done to better seal the edges of the cathodes and provide homogenous pressure onto the cathode to reduce contact problems.

The cell was either operated with ambient air or pure oxygen. The gas flow was led to a 3 cm high water column and set to 1 bubble per second. Compared to fuel cells, batteries exhibit much lower currents densities and therefore a homogenous gas distribution can be achieved more easily. As a consequence, no forced convection by pumps is necessary, however, a sufficient gas supply is mandatory which was accomplished by providing enough gas inlets/holes. In our case up to four inlets were integrated.

Due to the fact, that potential can be measured more precisely than current, a core element of the segmentation diagnostics is the current to potential conversion. In this work it was carried out by using two different kinds of operational amplifiers (OpAmp) circuits. The transimpedance amplifier (TIV) and the instrumental amplifier (InAmp). The following part is divided into the description of the setup, the functionality and the main advantages and disadvantages of the certain amplifier circuits.

Transimpedance amplifier (TIV):

The TIV, depicted in Fig. 2 is an inverted amplification circuit, based on the signal source being connected to the inverted input. U_E describes the input and U_A the output signal. All existing resistances before the inverted input are condensed in an equivalent resistor R₁. R₂ is the feedback resistor. The negative input is called inverted and the positive none inverted. ┴ symbolizes ground potential (GND). The bipolar potential source of the operational amplifier is marked by V_- and V₊.

Zoom In Zoom Out Reset image size

The non-inverted input is connected to ground (GND). As a consequence of the non-inverted input being connected to GND and the input potentials are set to the same potential both input signals and the output potential U_A share the same potential. An incoming potential coming from the input (U_E) increases the potential according to the ohmic law. The current flow through the feedback resistor R₂ is connected to an internal circuit of the amplifier and with this is connected to the negative power source. The amplifier evens out the amplified potential by supporting a current flow over the feedback wiring to V_-. As a consequent there is a potential drop at the amplifier output U_A. The resistor R₂ and the new output potential U_A results in an equivalent current compared to the input current. These currents balance each other out at the inverted input. The amplifier maintains the potential between the potential source and the inverted input. Since the potential drop of R₂ is controlled by the operational amplifier, the signal source is not being influenced by R₂ which avoids errors commonly present with shunt resitors setups.

The amplification is calculated with the following formula:

$$\begin{eqnarray}&&{U}_{A}=-{U}_{E}\cdot \displaystyle \frac{{R}_{2}}{{R}_{1}}\end{eqnarray} \tag{ 1 }$$

Instrumentation amplifier (InAmp):

The InAmp is a differential amplifier (Fig. 3). U_E describes the input and U_A the output signal. R_S describes the shunt resistor and R_V the setting resistor. ┴ symbolizes ground potential (GND). The bipolar potential source of the operational amplifier is marked by V_- and V₊.

Zoom In Zoom Out Reset image size

The current from the signal source is flowing through the shunt resistor R_S. There is a potential drop over the resistor dependent on the incoming current and the resistor value, which is measured by the OpAmp. The potential then is amplified by a factor, which is determined by the setting resistor R_V. The output potential is raised respectively. Table I gives an summary and comparison of the TIV and InAmp circuit.

The above described characteristics of both circuits determine the sources for errors. These errors were calculated by error propagation (see supporting information) and showed an approximately 10 times smaller error for the TIV. This is mostly due to the shunt and variable resistors. In the following experiments the TIV is mainly used because of the lower errors.

A Zahner electric Im6 equipped with 4 PAD4 (Parallel Analog-Digital 4) cards for parallel voltage measurements was used to record the current distributions. During charge and discharge constant currents were applied and the sampling rate was set to 2 s.

For the validation of the setup, an experiment with a variation of the electrolyte level which leads to a varying amount of electrochemically active electrode segments was performed and the current distribution was measured. Furthermore, two different types of operational amplifier circuits were compared in their ability to measure and resolve currents from the segmented current collector. The aim was to identify an electronic design with signal of high quality and resolution. For a description of the circuits with the corresponding features, see section segmented cell techniqu.

Figure 4 shows the comparison of the transimpedance amplifier (TIV) and the instrumentation amplifier (InAmp) circuit board connected to the cell for an increasing electrolyte level. 16 electrically isolated silver electrodes were used in a 4 × 4 arrangement. The electrolyte level was raised 4 times (15 mm, 30 mm, 45 mm, 65 mm), with each level adjusted to fully contact the next row of electrodes and the current distribution was recorded at a fixed current of 160 mA. Since the electrochemical reaction can only take place where electrolyte is in contact with the electrodes, the applied current will flow through the wetted electrode segments. This results in locally high current densities (depicted in yellow) for low electrolyte levels, as the total current is fixed for all measurements. When the electrolyte level is raised up to the next row of electrodes the contacted segments become active leading to a more homogeneous current distribution. At full contact deep blue regions vanish and the overall current distribution is homogeneous with modest local current densities. As expected both circuit boards show this behavior demonstrating the functionality of the setup.

Zoom In Zoom Out Reset image size

While the TIV and the InAmp circuit show comparable results for an electrolyte level of 15 mm and 65 mm, both electrolyte levels of 30 mm and 45 mm show some deviations. The formation of a homogeneous current distribution seems to be better for the TIV circuit. Especially in the center region of the InAmp circuit measurement only low currents were recorded. The use of 16 single electrodes placed in the PTFE frame could inflict contact problems since during mounting the electrodes it was not possible to place them completely in plane with the PTFE frame. This can be seen for the InAmp circuit at segment 6 (cf Fig. 1 red circle), which always shows very low currents, e.g. for the fully filled cell segment 6 shows around 6 mA compared to 10 mA at the other electrodes. In further studies this electrode arrangement consisting of 16 single electrodes will be replaced by one electrode which covers the whole measured area. In this respect it has to be mentioned that for all segmented measurement techniques the local resolution is determined by the in-plane electric conductivity which counteracts the heterogeneity of current distributions. Since silver exhibits very high electronic conductivity we needed to have an insulating frame. The following results in the following figures were recorded with the TIV set up.

To further validate the setup and demonstrate its capabilities, the effect of gas supply location was investigated. To this end, three gas ports were sealed (Fig. 5 red crosses), only the gas port at the lower right was left open (Fig. 5 green circle). Figure 5 shows the current distribution at the start of the measurement (left figure) and the normalized current distribution at the end of the measurement after reaching 0.5 V vs Zn (right figure). The normalization was done by subtracting the initial current distribution from the current distribution of interest and was done to better identify changes between start and operation and to mitigate small deviations in the initial state most probably ascribed to contact problems. During the measurement a clear shift of current to the lower right corner with the open gas port can be identified while the rest of the electrode shows a decrease of current indicating the depletion of oxygen and therefore lower reaction rates. The three corners with closed gas inlets still show higher current densities than the center of the electrode. This could be explained by diffusion of oxygen through the sealing.

Zoom In Zoom Out Reset image size

To investigate the influence of the flow rate in more detail, the flow was increased 3-fold from 0.065 ml s⁻¹ increased to 0.327 ml s⁻¹ and set back to the initial 0.065 ml s⁻¹. At each rate the flow rate was maintained for 2 min. The flow direction was adjusted by using only one gas port as inlet and another one as outlet. Accordingly, the gas flowed either from the lower right (lr) to the upper right (ur), or from the upper left (ul) to the lower left (ll), respectively. Both tests were done on the same setup including the same electrodes. First the gas flow was set to the right side (lr→ur) and afterwards to the left side (ul→ll). The results are shown in Fig. 6.

Zoom In Zoom Out Reset image size

The current distribution at t = 0 shows an acceptable homogeneity. For the gas flow from lower right (lr) to the upper left (ul), segment 16 (upper left corner) shows a slightly higher current than the other segments which is ascribed to the position of the inlet (Fig. 6 upper line). This effect increases with the increase of the flow rate as can be seen in the middle figures for high gas flow rate. For the flow on the right side the reaction shifts to the left, while for flow on the left side the reaction shifts to the right. In our setup the gas flowed through the GDL which inflicts resistance to the gas flow and therefore led to increased gas pressures for higher flow rates. The increased gas pressure shifts the location of the three-phase boundary away from the gas towards the electrolyte side of the GDE, thereby reducing the electrochemically active surface area. After reducing the gas flow again to the initial value, the current distribution levels off and exhibits a more homogeneous distribution again. This can be seen clearly for the ul→ll test while for the lr→ur test the initial state was not achieved again. For all segments a small drop in current density was observed after around 1 min (see Fig. S1 in supp. Information (available online at stacks.iop.org/JES/168/120530/mmedia)). This drop might explain the deviation from the initial state as the system might not have been in equilibrium state. This is because changes in current density also lead to changes in the surface coverage with reactants as well as changed electrolyte saturation of the surface due to changed electrowetting.

It has to be mentioned, that the pressure inside the porous system which builds up as a result of the change in the gas flow is more severe than what is commonly observed for membrane electrode assemblies (MEAs) in fuel cells. The cell used in this study did not have a flow field, but instead uses gas ports which lead the gas directly through the GDL. Flow fields could homogenize the gas pressure throughout the channel system, and reduce the pressure inside the porous structure as the gas would not have to be transported through the GDL to its full extent. An additional difference is the thickness of MEAs compared to our electrodes. That said, the presented technique can serve as a valuable tool in evaluating and optimizing cell design and flow-field structure in future work.

The observed current inhomogeneities cause heterogeneous aging. For fuel cells such degradation investigations were carried out by e.g. Durst et al. and showed that fuel starvation under the land compartment of the flow field leads to decreased degradation. ³⁴ This is because oxygen has to diffuse from areas with channels to areas without channels. In our case the forced heterogeneous gas distribution has the same effect. In areas with low concentration of reactant lower reaction rates are present, whereas in areas with sufficient supply of oxygen the current densities are high. These regions show faster degradation.

Inhomogeneities on the cathode side might also inflict inhomogeneities on the anode side. These inhomogeneities can be due to several reasons like gas flow and distribution, as it was shown above but also due to different penetration depths of electrolyte in the porous system. The wetting behavior affects the active surface inside the porous electrode. To investigate the influence of inhomogeneities on the anode, the battery was discharged to 0.5 V vs Zn. The current was set to 240 mA and the gas flow was set from both lower gas ports to both upper gas ports. The result is depicted in Fig. 7. The first row shows the normalized state at 2.78 h and was used to normalize the current distribution for better visualization of changes, as at this point the system was in a stable condition (the non-normalized distribution can be found in the supporting information Fig. S3). Row 2 shows the result at the first visible hole in the anode at 7.37 h. This is accompanied by a shift of current to the bottom of the cell where the anode is still intact. At 9.43 h the hole in the anode grew larger and the current shift to the bottom has advanced further. At the end of the test (11.5 h) the anode is almost completely dissolved and the cathode can be seen through the cell. A potential and current diagram over time is shown in the SI in Fig S4.

Zoom In Zoom Out Reset image size

The faster dissolution of the anode is due to higher currents at the top of the cell. Reason for this might be either the hydrostatic pressure of the electrolyte or the pressure induced by the gas flow. The influence of the gas flow was investigated in Fig. 5. As we used the low flow rate from bottom to top on both sides the influence of the gas flow rate onto the current distribution should be negligible. The hydrostatic pressure on the other hand inflicts higher electrolyte pressure at the bottom of the cell and therefore higher penetration depths. This reduces the active area and therefore has a similar effect than the high gas flow rate would show.

This inhomogeneous dissolution of the anode hinders high cycle life since the structure of the anode is disintegrated at some areas, while on others is still intact. During charging inhomogeneous currents inflict higher local currents resulting in dendrite growth, ³ which short circuits the battery.

Parker et al. proposed a structured zinc sponge consisting of zinc with interconnected pores in between. The capacity of such electrodes cannot be used completely since a backbone structure is needed to precipitate metallic zinc during charging back onto the structure. Inhomogeneous current distributions during discharge, like observed by us, would destroy such host structures resulting in reduced capacity and cycle life. ^3,35,36 The loss of usable zinc is accompanied by gravimetrical forces in vertical cells, like the one used by us. This is due to the reaction pathway where first zinc hydroxide is formed. After reaching the solubility limit zinc oxide formation takes place. Therefore, the current distribution should be optimized to accommodate for hydrostatic pressure influences on cathode side as well as gravimetrical forces on anode side.