Architecture-Based Control of Temperature Gradient-Driven Water Transport in Polymer Electrolyte Fuel Cells

A 4.8 cm² active area fuel cell was developed to facilitate high resolution neutron imaging with precise thermal boundary conditions. The cell has a narrow 9 mm × 53 mm active area to reduce geometric unsharpness and beam hardening effects. Coolant channels were directly behind the current collectors to provide precise thermal boundary conditions on the cell. Coolant temperature was controlled via circulating bath with ±0.2 °C accuracy. Additional details on the cell, along with information on reduction of the neutron images and appropriate corrections, can be found in Ref. 25.

To better understand the driving force behind water accumulation in the anode seen in previous work,²⁵ two flow field configurations were tested: the baseline asymmetric configuration developed in Ref. 26 and a new, symmetric flow configuration. This symmetric configuration uses the 1 mm cathode pitch from the asymmetric configuration for both the anode and cathode, as shown in Fig. 1. The symmetric and asymmetric flow fields consist of a triple serpentine flow pattern with 0.5 mm square channels. The symmetric pattern has 0.5 mm wide lands on both the anode and cathode whereas the asymmetric pattern has 1.5 mm wide anode lands and 0.5 mm wide cathode lands.

Zoom In Zoom Out Reset image size

The materials tested were the same as in previous work.²⁵ The DM used was Mitsubishi Rayon Corporation (MRC) U105 with 230 μm thickness, 6 wt% PTFE, and 30 μm MPL coating. W.L. Gore & Associates, Inc. Select Series 5720 18 μm thick membranes with 0.3 mg cm⁻² cathode loading and 0.05 mg cm⁻² anode loading were used. To understand the role of thermal conductivity of the DM on anode water accumulation, SGL 25BC was tested on a limited basis. MRC U105 has a thermal conductivity of 0.18 W m⁻¹-K^{−1 1} while SGL 25BC has a higher thermal conductivity of 0.35 W m⁻¹ -K⁻¹.^1,2,9

Neutron imaging was conducted at the BT-2 Neutron Imaging Facility at the NIST Center for Neutron Research located in Gaithersburg, MD. Imaging was achieved with the 15 μm effective resolution microchannel plate detector. The Neutron Imaging Facility has standard fuel cell testing equipment (Micro Instrument Corp hardware with LabView software) that allows precise control of gas flow rates, gas composition, gas and cell temperatures, gas humidification, inlet and outlet pressure, current, and voltage.

Each test case was run for 5 min to stabilize and then imaged for 20 min at steady state. The long image length was required to achieve sufficient neutron counts for each image; additionally, stable operating conditions (e.g. current, voltage, temperature, humidity) were verified during the exposure time. Detailed information regarding reduction of the neutron images to water thickness and conversion to saturation can be found in Ref. 25. In short, liquid water attenuates neutrons passing through the test cell; by comparing a neutron radiograph with no test cell, a radiograph with a dry test cell, and a radiograph of a saturated/operating test cell, water content can be measured. Porosity data for the MRC diffusion media was obtained from Ref. 27.

Test conditions were the same as described in Ref. 25 to allow for direct comparison of the symmetric and asymmetric cell configurations. Tested variables include current density, cell temperature, inlet relative humidity, exhaust pressure, and water balance. Flow rate was held constant across all test conditions regardless of current density with an equivalent stoichiometric ratio of 2∣2 (anode∣cathode) at 1.2 A cm⁻². Inlet relative humidity was held at a constant 95∣95%RH (anode∣cathode) during exhaust pressure testing while exhaust pressure was held constant at 150∣150 kPa(a) (anode∣cathode) during relative humidity testing. An overview of conditions tested is given in Table I. Cell water balance was measured for select test cases by collecting all exit water with desiccants. By collecting exit water in this manner, the overall net water flux could be verified; insight regarding the processes that contribute to that measured net water flux is the subject of the model formulation described next.

Two-dimensional temperature distributions were calculated for the symmetric and asymmetric configurations for dry and wet cases using COMSOL 4.3a. A detailed description of how thermal conductivity is adjusted for saturation is given in Ref. 28. Heat generation was determined using Eq. 1.²

$$\begin{eqnarray}&&q\prime\prime\prime =\displaystyle \frac{i}{{t}_{cl}}\left[\left(\displaystyle \frac{T{\rm{\Delta }}S}{nF}\right)+\eta \right]\end{eqnarray} \tag{ 1 }$$

Where i is current density, t_cl is the catalyst layer thickness, T is temperature, ΔS is the half-cell entropy change, n is the number of electrons per mole of reactant, F is the Faraday constant, and η is the overpotential. The entropy change for the cathode reaction is −326.36 J mol⁻¹-K⁻¹ and 0.104 J mol⁻¹-K⁻¹ for the anode. Joule heating is considered for the membrane. The values for the variables in Eq. 1 and used in the computational model are given in Table II.

The 2-D temperature profiles were generated and then used to calculate the PCI flux across the anode and cathode diffusion media. Temperature profiles are taken along the DM∣CL and DM∣channel interfaces for both DM and are used to calculate water vapor concentration gradients across each DM via Eq. 2.

$$\begin{eqnarray}&&C=\displaystyle \frac{{P}_{sat}}{RT}\end{eqnarray} \tag{ 2 }$$

With concentration as a function of temperature, Fick's Law of Diffusion (Eq. 3) was used to calculate the maximum flux across each DM for each node along the interface.

$$\begin{eqnarray}&&{J}_{PCI}={D}_{eff}\displaystyle \frac{\partial C}{\partial x}\end{eqnarray} \tag{ 3 }$$

To determine which half of the cell had the dominant PCI flux, a new parameter was developed. This parameter is called the thermal transport number (TTN) and represents the normalized cathode PCI flux with respect to generated water subtracted from the normalized anode PCI flux, as shown in Eq. 4.

$$\begin{eqnarray}&&TTN={\left(\displaystyle \frac{{J}_{PCI}}{{J}_{gen}}\right)}_{anode}-{\left(\displaystyle \frac{{J}_{PCI}}{{J}_{gen}}\right)}_{cathode}\end{eqnarray} \tag{ 4 }$$

This equation represents the maximum thermal transport driving force for water transport. A positive TTN value represents a preference for transport away from the anode while a negative number implies a preference for transport away from the cathode. Assuming the anode and cathode sides of the membrane have the same water content, and the membrane has zero resistance for water transport, a positive TTN would draw water to the anode from the cathode layer by creating a diffusion gradient across the membrane.

Water thickness data measured with neutron imaging was converted to local saturation values for each condition. Figure 2 shows that water saturation varies significantly between the two configurations at 40 °C and over a range of fuel cell exhaust pressures. The asymmetric configuration has large sensitivity to the imposed gradient with a large range of anode water content. Maximum saturation varies from approximately 15% for the 150∣100 kPa case to 60% for the 100∣150kPa case. This sensitivity is not seen in the symmetric configuration. Maximum anode water content does not scale 1:1 with land width as the asymmetric configuration has 3 times the land width but only double the water content. The cathode saturation profiles for both configurations follow similar trends with matching boundary saturation values at the MPL and the flow field. The increase in peak water content from 30% for asymmetric to 35% for the symmetric configuration indicates improved water removal from the cathode. This would agree well with the reduction in anode water content.

Zoom In Zoom Out Reset image size

For a fuel cell temperature of 60 °C, exhaust pressure saturation profiles are compared for both configurations at 1.2 A cm⁻². Similar trends are seen at 60 °C (Fig. 3) as seen at 40 °C with small variation between symmetric configurations and large water accumulations in the asymmetric configurations. Large variation in the asymmetric water content between cases is not seen at 60 °C due to the critical maximum saturation level being reached. This is evident by the fact that the 100∣150 kPa and 150∣150 kPa [anode∣cathode] profiles overlay each other. At this point liquid water connects across the lands to facilitate drainage, thus limiting increases in saturation beyond this point. Cathode peak saturation increases for the symmetric configuration from approximately 30% to 35% due to increased water removal on the cathode side. The boundary conditions for the cathode remain consistent between configurations even with increased peak saturation for the symmetric configuration. This is expected as the DM boundaries (MPL and flow field) are identical between configurations and this behavior should follow between the two.

Zoom In Zoom Out Reset image size

The reduction in anode water content in the symmetric configuration indicates both a reduction in back diffusion of water from the cathode to the anode and reduced storage due to a change in critical saturation level. Cell water balance was measured for all pressure conditions for both configurations at 60 °C to determine the reduction in back diffusion from the cathode to anode. It can be seen for all cases in Fig. 4 that the symmetric configuration has a larger cathode removal fraction of generated water than the asymmetric configuration. This implies that the asymmetric configuration has greater back diffusion to the anode than the symmetric configuration. The large anode lands of the asymmetric configuration provide large cool regions that drive thermal transport from the cathode to the anode. To further understand the interplay between land geometry and DM thermal properties on the water balance of the cell, an asymmetric cell was built with SGL 25BC diffusion media and imaged. SGL 25BC has similar thickness, porosity, and PTFE content while having double thermal conductivity of the MRC. Figure 5 shows that the SGL asymmetric configuration has significantly lower water content than the MRC asymmetric configuration. By increasing the DM thermal conductivity, the peak temperature of the cell decreased from 74.2 °C to 67.7 °C which significantly reduces the thermal driving force for PCI flux across the cell.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

At 60 °C, inlet gas stream relative humidity was varied for both the anode and cathode to determine the influence of concentration-driven flow across the cell. Figure 6 shows that the symmetric configuration is invariant with inlet relative humidity while the asymmetric configuration exhibits high sensitivity of saturation depending on humidity levels. As with the pressure conditions, cathode boundary interfaces are similar between configurations, indicating no change in boundary interaction due to increased or decreased levels of water transport. The symmetric configuration has decreased anode water content while the cathode has a higher peak saturation that shows a larger flux of water is transported out via the cathode than in the asymmetric configuration. Cathode variability in the asymmetric configuration follows anode saturation levels with lower cathode saturation for the two dryer anode cases.

Zoom In Zoom Out Reset image size

Water content is significantly reduced for both configurations when the temperature is raised to 80 °C as shown in Figs. 7 and 8. The higher temperature facilitates greater vapor transport and reduced levels of condensation within the DM. As seen at the lower temperatures, the asymmetric configuration has higher anode saturation than the symmetric configuration for both the exhaust pressure conditions and inlet relative humidity conditions. A slight increase in cathode saturation is seen for the symmetric configuration; this shows increased water removal through the cathode. The decrease in water content for both configurations allowed all test conditions to run through the entire test duration. Operating voltages for all test conditions presented in this section are given for both configurations in Table III.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To better understand the driving force behind the large water accumulation in the asymmetric configuration, it is necessary to understand the influence of the large anode lands on the 2-D temperature profile. Heat transfer models in COMSOL are used to solve for the 2-D temperature distribution for a 2 mm long representative segment for both configurations. This domain represents the distance required for symmetric boundary conditions for the asymmetric configuration. The TTN is determined along every node along the 2 mm domain which is plotted in Fig. 9. This figure shows the TTN profiles for the 150∣150 kPa cases for each configuration. A positive TTN indicates a preference towards anode-driven water balance while a negative number indicates cathode-driven water balance. Anode and cathode lands for both configurations are overlaid with the plot to show how they influence the TTN.

Zoom In Zoom Out Reset image size

It is evident in Fig. 9, the symmetric configuration favors cathode transport while the asymmetric configuration has a large positive spike in the center of the domain that aligns with the center of the large anode land. This spike is due to the insulating effect the cathode channel has in the center of the domain which results in a low CL-to-channel temperature gradient on the cathode side. The large thermal driving force towards the anode in the asymmetric configuration allows pumping action to draw water to the anode where it collects over the large lands. Transport to the anode is facilitated by the fact that cationic membranes such as those used here preferentially pump water from hot to cold side via thermo-osmosis.^7,29,30 Variation in the TTN for the symmetric configuration is small compared to the asymmetric configuration. This indicates that water accumulation should be lower for the symmetric configuration which agrees well with the neutron imaging data.

The TTN plot can be used to describe the variation in the asymmetric configuration for the exhaust pressure tests as shown in Fig. 10. It can be seen for both configurations that the 100∣150 kPa exhaust pressure case has the strongest preference towards the anode. The integrated value of the 100∣150 kPa case for the symmetric configuration is only 65% of the integrated TTN value for the asymmetric configuration. The lower saturation seen in the symmetric 100∣150 kPa case when compared to the 150∣150 kPa asymmetric case indicates that land width strongly influences water storage. The lag seen in water accumulation for the asymmetric configuration in the 150∣100 kPa case is due to the larger preference towards the cathode compared to the other two cases. Water flux across the membrane is reduced in this condition requiring additional time to accumulate water in the anode DM.

Zoom In Zoom Out Reset image size

As has been shown in previous work,¹ saturation strongly influences thermal conductivity; similarly, saturation influences the temperature gradient based driving force for thermally-driven transport. Increased thermal conductivity can reduce peak temperature in the symmetric configuration by 8.1 °C and the asymmetric by 8.4 °C at 1.2 A cm⁻². Temperature profiles were calculated to determine wet TTN values using the saturation values from the neutron images to determine wet thermal conductivity for both configurations. The TTN values for both configurations for wet conditions at 150∣150 kPa are given in Fig. 11. large attenuation can be seen for maximum TTN due to the reduction in driving force and increased diffusion resistance due to liquid water accumulation. This shows that thermally-driven water transport continuously changes during fuel cell operation, with the highest driving force available when operating with no liquid water in the cell. The value of TTN will slowly decay as water builds up in the cell by reducing peak temperature and increasing diffusion resistance until a balance occurs in the cell. If a large transient draining event occurs during steady current density operation, the water balance can swing back towards the anode resulting in a rapid refilling of the anode DM. For full scale fuel cells operating in counter-flow, rapid anode water accumulation can occur close to the anode inlet as water buildup near the cathode exit would reduce water transport through the cathode and force it over to the anode.

Zoom In Zoom Out Reset image size

Vapor carrying capacity of the gases increases significantly (and in a highly non-linear fashion) with increasing temperature; it is expected that the maximum TTN value would increase with operational temperature. A fourfold increase in TTN is seen in Fig. 12 as temperature increases from 40 °C to 80 °C for the asymmetric configuration operating at the 150∣150 kPa test case. The increase for the symmetric configuration towards cathode preference is more subdued due to the more identical anode and cathode temperature gradients. This shows that during cold startup, thermally-driven transport would be a smaller percentage of the entire transport makeup but would become the dominant driving force at normal operating temperatures.

Zoom In Zoom Out Reset image size

It is important to understand how current density affects thermally-driven transport as this affects operational voltage and heat generation within the cell. For the conditions in this work, saturation increases with increasing current density as shown in Fig. 13. little water buildup is seen at the lowest current densities due to the constant flow rates used, resulting in stoichiometric ratios in the range of 20 for 0.1 A cm⁻². It is noted here that the symmetric configuration could not reach a stable 1.5 A cm⁻²; mass transport limitation resulted in cell shutdown for the relatively saturated cathode. Figure 14 shows the TTN plots for both symmetric and asymmetric configurations for five current densities increasing from 0.1 A cm⁻² to 1.5 A cm⁻² at 60 °C. For the symmetric configuration, the TTN indicates PCI flux towards the cathode at all current densities, leading to greater cathode saturation seen in Fig. 13. Generally, it can be seen that the operational current density has only a minor influence on the TTN value as the PCI flux is normalized with the water generation for the given current density. Due to the saturation remaining low at the 0.1 and 0.4 A cm⁻² current densities, thermally-driven flow remains constant for the duration of the test whereas the higher current densities decrease in TTN value with increasing saturation. Thermally-driven flow can be the dominant transport mode across all current densities for the asymmetric configuration. This makes thermal design an important consideration even for fuel cells that operate only at low current densities.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size