Fuel Cells for Space Applications

Fuel cells are electrochemical devices that combine hydrogen fuel and oxygen to produce electricity, with heat and water being the only byproducts. They operate without combustion and are therefore virtually pollution free. Fuel cells can operate at much higher efficiencies than internal combustion engines, since the fuel is converted directly into electricity, and the device has no moving parts. A proton exchange membrane (PEM) fuel cell consists of an anode, an electrolyte (proton exchange) membrane in the center, and a cathode. As hydrogen flows into the fuel cell anode, a platinum coating on the anode facilitates separation of the hydrogen gas into protons and electrons. The electrolyte membrane in the center allows only the protons to pass through the membrane towards the cathode. Since the electrons cannot pass through this membrane, they flow as an electric current through the external circuit, which provides electric power to the load. As oxygen flows into the cathode region, another platinum coating causes the oxygen, protons, and electrons to combine, producing water and heat.

Individual fuel cells can be stacked, and the number of fuel cells determines the total voltage V, while the surface area of each cell determines the total current I. The product IV is the total power generated.

NASA used fuel cells to provide onboard power for the Gemini and Apollo spacecraft in the 1960s and, more recently, the Space Shuttle. There is a need for noninvasive methods of characterizing the dynamics of electrochemical PEM fuel cell reactions, as well as imaging any flaws in the membrane that may exist prior to failure.

Technical Plan and Equipment
Electric currents generated in PEM fuel cells give rise to magnetic fields that can be measured with a SQUID or fluxgate magnetometer. The magnetic field strength is directly proportional to the total cell current while the spatial field map can be used to determine variations in the underlying current distribution due to flaws in the PEM. Current fluctuations in the fuel cell give rise to variations in magnetic field that can be analyzed in the time or frequency domain. Magnetic fields have complex spatial, frequency and time dependencies governed by the cell geometry, membrane health, and operating conditions.

We have previously demonstrated the use of SQUIDs to measure the field produced by various electrochemical reactions. An analysis of noise during reactions of metals with liquid electrolytes has revealed that electrochemical noise phenomena is related to a class of physical systems said to exhibit Self Organized Criticality (SOC).¹ Such a connection between electrochemical noise and SOC may further our understanding of corrosion reactions and current fluctuations in batteries and fuel cells. We also demonstrate electrochemical noise analysis as a noninvasive tool for the NDT of fuel cells.

FY2000-01 activities included the integration of a fuel cell test station with the SQUID measurement system as well as the simultaneous data acquisition and motion control of an XY scanning table using the Labview software. Figure 1 shows the experimental setup for spatial field mapping using a High-T_c SQUID magnetometer. The SQUID is cooled to 77 K in a G-10 fiberglass liquid nitrogen dewar. A cryogenic cable connects the SQUID to the Flux Locked Loop (FLL) which linearizes the SQUID voltage as a function of magnetic field. Data are transferred from the FLL to the SQUID controller via fiber optic cable. The SQUID and dewar assembly are suspended over the fuel cell by a gantry that can be raised or lowered with a crank shaft. The fuel cell is positioned on an X-Y scanning table with a nine-inch full range in each direction. The scanning speed, step size, and grid dimensions are computer controlled using the Labview software package. The SQUID output is low pass filtered and output to an AD converter using the auxiliary input of a lock-in amplifier. Data are simultaneously acquired during the C-scan using the same Labview program.

Figure 1. Liquid nitrogen dewar (top left hand corner) suspended above a singe PEM fuel cell mounted to an XY scanning table. The Fuel cell test station (Globetech) is located to the right.

The fuel cell and test station used in this study is manufactured by Globe Tech. The single cell Membrane Electrode Assembly (MEA) consists of two graphite blocks that surround the PEM. Gas flow channels facing the PEM are machined into the graphite blocks. Copper electrodes, in turn, surround the graphite blocks. The entire assembly is held together with brass bolts so that all of the fuel cell parts are nonmagnetic. H₂ and O₂ gasses are supplied to the cell from the gas-flow controller. The gasses can be humidified or supplied to the cell in dry form. Current leads connect the fuel cell to a variable load bank. The experiment conducted with the fluxgate magnetometer is similar to the SQUID setup with fluxgate voltage supplied to the AD converter.

Experimental Activity
Experiments were conducted using both SQUID and fluxgate magnetometers to measure the magnetic fields produced by currents flowing in the fuel cell. DC magnetic field maps were recorded with fixed magnetometers located above the membrane electrode assembly. The signal of interest in these measurements is the magnetic field resulting from currents that flow in the PEM. To measure these fields several sources of electromagnetic interference must be eliminated. These include the Earth's magnetic field, 60 cycle power-line noise, the field produced by stray currents in the electrodes and input leads, and nearby ferromagnetic objects such as the stepper motors. Low pass filtering is used to attenuate noise sources above 5Hz. The fuel cell is elevated on a nonmagnetic platform elevated 12 inches above the XY positioning table. A superconducting tube surrounding the SQUID serves to reduce ambient noise from distant sources. A reference scan is made without the fuel cell and subtracted from the raw data to remove the signal contribution of nearby magnetic sources. Further noise reduction is performed in the Labview program by taking 20 data points at each position and averaging.

Figure 2 shows the results of a one-dimensional scan over the fuel cell. The raw data-reference and subtracted output-are shown for comparison. The difference between the raw data and the reference scan is solely due to currents flowing in the fuel cell. Figure 3 shows the magnetic field map obtained by a 2D scan over the fuel cell after the reference subtraction. In these scans current leads are input into one the side of the fuel cell. Field maps obtained in this configuration are qualitatively similar to the field generated by a current loop consisting of a large peak centered over the loop. The fuel cell is located in the center of the scanning region in Fig. 3, however, the field peak is shifted to one side because of asymmetrical currents in the cell resulting from the position of the input leads.

The total current that flows through the membrane also flows through the electrodes and the input leads. Each of these currents generates a field that is detected by the magnetometer. Therefore, the contribution to the total field generated by the electrodes and the input leads must be cancelled. Twisting the input leads effectively nulls the field contribution of each lead. If the current leads are connected directly beneath the fuel cell, the electrode currents flow primarily in the vertical, or z-direction. Because magnetic field is always perpendicular to the direction of current flow, a vertically oriented magnetometer will only detect field generated by the horizontally directed membrane current.

Figure 4 shows an ac field map obtained using a fluxgate magnetometer. This experiment was conducted using a current injection technique where current is supplied to the cell from an ac source. The fluxgate voltage (proportional to the total field) is output to a lock-in amplifier referenced to the signal generator. The lock-in amplifier serves as a band pass filter detecting only the ac field due to currents flowing in the fuel cell. This technique has the advantage of enabling inspection of the cell without supplying fuel gasses, and also eliminates the need for a reference scan. Here, current leads are supplied from beneath the cell so that the measured field is due solely to currents flowing in the membrane. Field maps obtained in this configuration are qualitatively similar to the field generated by a current dipole consisting of both a positive and negative peak. Flaws in the membrane will give rise to variations in this current dipole membrane field.

Figure 4. 2-D ac magnetic field recorded scanning a fuel cell over the fuel cell. The magnetic field is due solely to currents flowing in the PEM.

Temporal variations in magnetic field (or current density) may result from degradation or poisoning of the membrane, electrochemical noise, and flooding in the cell. We have studied time changing fields generated by the fuel cell under various operating conditions. Figure 5 shows four time series recorded with the fluxgate located at a fixed position above the cell. In this experiment both fuel gasses are humidified. Each time series corresponds to a different operating voltage, as indicated in the figure. The initial peak in these recordings is probably due to discharging of the electric double layer when a load is applied to the cell. The resulting magnetic field variations are due to flooding of the gas flow channels next to the membrane.

Figure 5. Magnetic time series resulting from current fluctuations in the fuel cell.

Frequency domain electrochemical noise measurements are made connecting the fuel cell directly to an FFT Spectrum analyzer. The method is rapid, noninvasive, and can be applied to individual cells in a stack. Figure 6(a) shows the FFT of open circuit voltage fluctuations from 0.1 to 100Hz under nominal operating conditions (bottom curve) and during oxygen crossover (top curve). The open circuit voltage spectrum is qualitatively different than the spectrum recorded during crossover leakage. Figure 6(b) shows spectra of voltage fluctuations recorded both in an open circuit and with load. A reference measurement is also made with a cold fuel cell (bottom curve). Higher noise levels are observed in an open circuit mode or with zero cell current.

References
¹J. R. Claycomb, K. E. Bassler, J. H. Miller, Jr, M. Nersesyan, and D. Luss. "Avalanche Behavior in the Dynamics of Chemical Reactions," Phys. Rev. Lett. 87 (2001): 178303-1-4.

Publications
Claycomb, J. R., M. Nersesyan, D. Luss and J. H. Miller, Jr. "SQUID Detection of Magnetic Fields Produced by Chemical Reactions," IEEE Trans. on Applied Superconductivity 11.1 (2001): 863-66.
Claycomb, J. R., K. E. Bassler, J. H. Miller, Jr, M. Nersesyan, and D. Luss. "Avalanche Behavior in the Dynamics of Chemical Reactions," Phys. Rev. Lett. 87 (2001): 178303-1-4.
Lobera-Serrano, J. A., J. R. Claycomb, J. H. Miller, Jr., and K. Salama. "Hybrid Double-D Sheet-Inducer for SQUID-based NDT," IEEE Trans. on Applied Superconductivity 11.1 (2001): 1283-86.
Nersesyan, M. D., J R. Claycomb, J. T. Ritchie, J. H. Miller, Jr., and D. Luss. "Magnetic Fields Produced by the Combustion of Metals in Oxygen," Combustion Science and Technology 169 (2001): 89-106.
Nersesyan, M. D., J. R. Claycomb, J. T. Ritchie, J. H. Miller, Jr., J. T. Richardson, and D. Luss. "Electric and Magnetic Fields Generated by SHS," J. Material Synthesis and Processing 9.2 (2001): 63-72.

Theses or student reports
Yarbourgh, R. A. Student presentation.
Gogoshin, G. Student presentation.

Presentations
Claycomb, J. R., M. Nersesyan, D. Luss, and J. H. Miller, Jr. "SQUID Detection of Magnetic Fields Produced by Chemical Reactions," Applied Superconductivity Conf., Virginia Beach, VA, Aug. 2000.
Lobera-Serrano, J. A., J. R. Claycomb, J. H. Miller, Jr., and K. Salama. "Hybrid Double-D Sheet Inducer for SQUID-based Nondestructive Testing," Applied Superconductivity Conf., Virginia Beach, VA, Aug. 2000.

Funding and proposals
"Quantum Transport Phenomena in Charge Density Waves." THECB-ARP, Jan. 1, 2000-Aug. 31, 2002, $159,000.
"Scanning SQUID Microscope for Non-destructive Imaging of Subsurface Defects." THECB-ATP, Sept. 1, 2000-Aug. 31, 2002, $133,400
"Applications of High-T_c Superconducting Films and Devices." TcSUH, Sept. 1, 2000-Aug. 31, 2002, $40,000.
"Novel Applications of High-T_c Super-conducting Sensors." Welch Foundation, June 1, 2001-May 31, 2004, $150,000.

	Figure 2. Magnetic field recorded scanning a High-T_c SQUID over the fuel cell. The reference scan is subtracted from the raw data to obtain the field produced by currents in the fuel cell.
	Figure 3. 2-D magnetic field recorded scanning a High-T_c SQUID over the fuel cell (after reference subtraction).