On-Board Control System of Water Content inside FCV Stack by Electrochemical Impedance Spectroscopy

Wednesday, 4 October 2017: 10:40
National Harbor 3 (Gaylord National Resort and Convention Center)
C. Mizutani, M. Shiozawa (SOKEN, INC.), T. Maruo, and S. Aso (TOYOTA MOTOR CORPORATION)
Water content inside FCV stack must be controlled appropriately to reduce waste of energy while gas purge. Humidification by generated water is indispensable because ohmic overpotential increases as membrane dries up. However, gas transportation is blocked when water accumulates inside stack with excessive water generation. Accordingly, water content inside stack must be kept optimum during operation. Also water control is a significant issue for FCV operation below zero. In addition to liquid water evaporation, excessive water is removed at the water trapper during a drive, but water content does not decrease after the vehicle stops. Once the stack is cooled at night, it is concerned that accumulated water freezes and blocks gas transport at the next start-up. After the stop, it is possible to reduce water content by shut-down purge. There is, however a problem that the sound noise and energy waste by purge is significant. If it is possible to reduce water content not only during driving but also after the stop, purge time would be shortened and the noise and energy waste above can be also reduced. Thus, water content inside stack should be maintained appropriately during both driving and shut-down purge.

If we grasp water content inside stack and control it in appropriate quantities during a run, we can decrease water content at shut-down and energy and gas consumption which is required for purge. However, water content inside stack cannot be measured directly while operation. Therefore, a new sensing method is desired that water content is calculated from parameters available in car.

Accordingly, we focused on relation between oxygen diffusion resistance and water content and between oxygen diffusion resistance and two different frequencies impedance and suggested a detecting method to find water content from oxygen diffusion resistance. In this method, it was assumed that oxygen diffusion resistance in catalyst layer can be divided into two components: Rion caused by oxygen gas transport in solid and Rvoid caused by vacant space in catalyst. We obtained oxygen diffusion resistance by calculating them from 200 Hz and 20 Hz two-frequency impedance. If two-frequency superimposed impedance becomes obvious, we can estimate water content in stack. In this matter, we established detecting method to seek water content during driving from impedance measurement.

First of all, we obtained oxygen diffusion resistance at cathode catalyst layer. Oxygen diffusion resistance Rtotal was assumed to be the sum of gas diffusion resistance due to an obstacle of liquid water, Rvoid and diffusion resistance concerned with oxygen transport inside ionomer, Rion. Rion can be calculated from a property of matter. We assumed an electrical equivalent circuit and Cole-Cole plot profile and found the diameter of the plot through 20 Hz and 200 Hz two-frequency superimposed impedance measurement geometrically. The diameter of Cole-Cole plot stands for mass transfer resistance, and this equals to the sum of resistance components of activation and concentration overpotential, Ract+Rdif. Concentration overpotential resistance was in terms of partial oxygen pressure PO2 and oxygen diffusion resistance Rtotal, and Rtotal was dissolved by concentration overpotential ηdif calculated with overpotential separation based on Butler-Volmer equation and the diameter Ract+Rdif. Rvoid could be taken Rion from Rtotal.

Moreover, we operated cell with measuring impedance for a certain time and measured water content after the operation on a plurality of various conditions. 20 Hz and 200 Hz two-frequency superimposed impedance was analyzed by FFT. Water content was found out by comparing stack’s weight before and after removal of cell fastening.

As a result, oxygen diffusion resistance increased in both cases of light and heavy water content from the relation between water content and Rtotal calculated by impedance measurement on a plurality of different conditions. When water content is heavy and flooding is supposed to occur, liquid water blocks gas transport inside voids, and Rvoid increases. Conversely, it is considered that Rion increases with a drop of water content inside ionomer when relative humidity is low and drying-up is supposed to occur. Therefore, Rtotal which is the sum of Rvoid and Rion showed such a downward profile against the variation of water content.

Moreover, Rvoid increased simply as water content increased. As mentioned above, it was considered that gas transport was blocked by liquid water in large quantities, and it was verified that Rvoid is an indicator of oxygen gas behavior due to liquid water inside voids at catalyst layer. Because this increment of Rvoid for water content corresponds our approximate expression, we concluded that water content in cell can be found out uniquely from Rvoid measured by impedance spectroscopy.