燃料电池气体扩散层微孔层内部液态水动态行为的时间分辨X射线CT成像［诊断测量二十］
Visualization of dynamic behavior of liquidwater in the microporous layer of polymer electrolyte fuel cell during waterinjection by time－resolved X－ray computed tomography
Satoshi YamaguchiSatoru KatoAkihiko KatoYoriko MatsuokaYasutaka NagaiTakahisa Suzuki
Abstract
Liquid water behavior in the microporouslayer （MPL） of a polymer electrolyte fuel cell （PEFC） was investigated using time－resolved X－ray computedmicro－tomography （CT） to elucidate the unsteady flow during flooding． Thewater was injected into an MPL laminated on top of the substrate of a gasdiffusion layer （GDL） to simulate the cathode in an operating PEFC． A CT scantime as high as 4．2 s was achieved，which enabled capture of the water motion in MPL pores with sizes of ca． 30 and 100 μm． Dynamic water transport from a hydrophobic MPL pore to thehydrophilic carbonaceous binder that bonds the substrate fibers wasobserved．
Fig． 1． Setup for the water injection X－rayCT measurement． （a） Structure of the injection water cell． （b） Schematicdiagram of the experiment setup． （c） SEM image of the MPL surface． （d） Porediameter distribution of the GDL measured by mercury－porosimeter （Pore Master60GT， Quantachrome Instruments）． The MPLhas pores with a wide range diameter from 0．005 μm to 150 μm．
Fig． 2． X－ray CT reconstructed 2D and 3Dimages of the MPLs． （a） Cross－sectional image of dry MPL． （b） Binary image of（a）． （c） Region 1 with 30 μm pores， indicated by c in （b）． （d） Region 2 with100 μm pores， indicated by d in （b）．
Fig． 3． 3D images of region 1 with smallpores （class 30 μm） at （a） t ＝ 0 s， （b） t ＝ 495 s， （c） t ＝ 468 s， （d） t ＝ 900s， and （e） t ＝ 1500 s． The water is shown together with the MPL in theleft－hand panels． The right－hand panels show the water translucently withoutthe MPL． The small pores spatiallyadjacent to one another accumulate water in a sequential manner．
Fig． 4． 3D images of region 2 with largepores （class 100 μm） at （a） t ＝ 0 s， （b） t ＝ 600 s， （c） t ＝ 900 s， （d） t ＝ 1035s， and （e） t ＝ 1044 s． The left－hand panels show volume rendering of the waterwith the MPL． Translucent images of water without the MPL are shown in theright－hand panels． An increase in watervolume and rapid reduction was observed in the large pores． Water recedesbefore it completely fills the large pores．
Fig． 5． Water saturation obtained fromFigs． 4 and 5． Water saturation simply increases in the small pores． Incontrast， the level of water saturation in the large pores oscillates． Thecircle and square plots correspond to the sequences shown in Figs． 4 and 5，respectively．
Fig． 6． Water behavior within a large pore．（a） Horizontal CT slice images of the large pore in region 2． The water dropletin the large pore shrinks in a short time between a1 and a2． （b） Horizontalslice CT images of 28 μm closer to the substrate than slice A． The yellowregion indicates the carbon binder of the substrate． Water localized in the large pore moves to the carbon binder area as thedroplet shrinks． （c） Vertical CT slice images containing horizontal slicepositions A and B shown by orange dashed lines．
Fig． 7． Schematic image of the slice imagepositions shown in Fig． 6．
Fig． 8． Grayscale level time evolution ofthe region R in Fig． 6（b）， of which the size is 9．75＊9．75 μm． Black circlescorrespond to the times in Fig． 6． The increase of image contrast indicateswater input into the pores of the carbon binder， between 1100 s and 1200 s
Fig． 9． Schematic diagram ofhydrophobic／hydrophilic contact model．
Conclusion
Water motion in hydrophobic MPL pores wasdirectly observed using a time－resolved X－ray CT technique． Pores with a diameter of approximately 30μm were gradually filled with water， one by one． In pores with a diameter ofapproximately 100 μm， water in the pores was absorbed by small hydrophilicpores in the binder before it filled the large pore． This observationimplies fast water transport occurs due to microscopic hydrophobic／ hydrophilicinterface． Modeling of this effect would facilitate the design of MPLs withboth hydrophobic and hydrophilic components．
