1－1．5V电位循环燃料电池膜电极阴极催化层沿流道方向、流道和脊部的局域碳腐蚀状态分析［失效分析其一］

Use of X－ray computed tomography forunderstanding localised， along－the－channel degradation of polymer electrolytefuel cells

Jennifer Hack

Lara Rasha

Patrick L． Cullen

Josh J． Bailey

Tobias P． Neville

Paul R． Shearing

Nigel P． Brandon

Dan J． L． Brett

Abstract

The need to understand the effects ofdegradation in polymer electrolyte fuel cells （PEFCs） has led to thedevelopment of in－situ and operando imaging studies of failure mechanismsoccurring in their constituent materials， but studies using X－ray computedtomography （X－ray CT） for imaging have focussed primarily on a single region ofthe PEFC flow channel． Whilst studies have shown local variation in degradationrates using electrochemical techniques， this work employs identical－locationX－ray CT imaging to elucidate the local degradation of a membrane electrodeassembly （MEA） at various locations along－the－channel of a serpentine flowfield． Using a carbon corrosion specific accelerated stress test （AST）， in－depthanalyses of the catalyst layers （CLs） and crack network allow forquantification of the material’s degradation at the inlet， middle and outletregions of the flow channel on the cathode side． Imaging of the regions wasdone after 0， 2000 and 5000 cycles and results show significant regionalvariation in the extent of degradation of the cathode CL． The order ofdegradation was found to be outlet ＞ middle ＞ inlet， with theoutlet CL found to be thinner than the middle and inlet regions， with a greaterextent of cracking． Furthermore， additional land and channel degradationeffects were probed， with the land regions found to be less degraded thanchannel regions． This work further highlights the need to understand anddevelop ASTs that can promote a more uniform degradation rate across an MEA．

Figure 1 Exploded CAD view of the cell  assembly， including graphite plates， gaskets and the MEA． Holes were milled  for compression screws， which screw through the integrated holder； b） xz  orthoslice of the middle region after 2000 cycles， showing the channel and  land and indicating the regions that were extracted for channel （red dashed  line） and land （red solid line） analyses； c） cropped grayscale orthoslice  （middle， 2000 cycles） used for image segmentation and d） the segmented image，  showing cathode CL （orange）， cathode cracks （blue）， anode CL （green） and  anode cracks （pink）． The white scalebar in c） and d） represents 200 μm and  the axes coordinates are the same as those in b）．

Figure 2 a） Polarisation curve and b）  in－situ ECSA values calculated from the hydrogen

adsorption peak of the cyclic  voltammograms．

工况是cycling between 1 and 1．5 V at 500 mV  s－1 for 5000 cycles

Figure 3 a） Macro－CT scan of the sample，  showing the serpentine flow channels． The GDL is seen through the flow  channel． The three ROIs have been indicated， with the colour of the bounding  box corresponding to the ROI， as shown in the volumes imaged using X－ray

micro－CT of the b） inlet， c） middle and  d） outlet．

ROI：regions of  interest感兴趣的区域

Table 1 Table showing the average CL  thickness for each location at the various extent of

cycling． Standard deviations， σ， are  shown in brackets next to each value．

Figure 4 Cathode catalyst layer for a－c），  inlet BOL， 2000 and 5000 cycles， respectively；

d－f） middle BOL， 2000 and 5000 cycles，  respectively； and g－i） outlet BOL， 2000 and 5000 cycles， respectively． Dashed  line represents the outline of the flow field， with area outside being the land．  Black scale bar represents 300 μm in each case． The colour bar indicated  represents the local thickness of the CL， with increasing thickness of CL  from green to red colour．

CCR： cathode－crack ratio阴极裂缝比率

ACR： anode－crack ratio阴极裂缝比率

Figure 5 a） CCR and b） ACR for all stages  of cycling for inlet （navy）， middle （green） and outlet （orange）； c） Crack  connectivity shown for the inlet （navy）， middle （green） and outlet （orange） regions；  d） Grayscale orthoslice （left） and segmented crack network （right） of the  middle region after 2000 cycles． Node regions and y－axis connectivity are  indicated by a red circle and dashed line， respectively． Here only two  examples of nodes and a single example of y－axis connectivity are indicated，  but there are many more not indicated that are picked up by the analysis． In  this image， each discreet isolated crack is represented by a label and these  are assigned different colours． e） Graph showing the number of nodes where  connectivity is equal to three and e） number of separate crack labels．

Figure 6 Histograms and TEM images of the  particle diameter distributions for a a，b） fresh

electrode and for the degraded electrodes  at c，d） inlet， e，f） middle and g，h） outlet． Data was collected for 300  particles for each sample． Scalebar represents 50 nm in each image，  （b，d，f，h）．

Figure 7 a） Images of the average crack  width for inlet （top）， middle （central） and outlet

（bottom） regions． The colourmap corresponds  to the crack width， with narrower cracks

represented by purple／red colour and a  yellow colour representing a thicker crack． Scalebars represent 100 μm； b）  average crack widths and c） average CCR for each land and channel region． In  b） and c）， the inlet， middle and outlet regions are represented by blue，  green and orange column colours， respectively， with lighter colours  representing land regions and darker colours representing channel regions．

Conclusion

In this work， X－ray CT has been used forin－depth investigation of along－the－channel degradation of a cathode CL using acarbon corrosion specific AST． Identical－location X－ray micro－CT had been usedto analyse the localized degradation of PEFCs across three regions of theflow－field during accelerated stress testing， namely the cathode gas inlet， amiddle region and the cathode gas outlet． A bespoke cell was designed fortesting， with a novel miniature design employing a serpentine flow－field designetched directly into the fuel cell end plate casing． Results of electrochemicaltesting showed a significant decrease in performance over the course of theAST， with a 27％ loss of ECSA after 5000 cycles． To understand themechanical degradation arising as a result of the AST， X－ray CT scans wereperformed after 0， 2000 and 5000 cycles of an AST at the three ROIs． In allthree regions， the cathode CL was found to degrade with increasing cycle number，primarily via formation and growth of crack networks． The anode did not degradeover the course of the AST． However， perhaps more significant was the observationthat the rate of degradation across the three regions of the MEA varied significantly；the inlet was found to degrade less than the middle region， with the outletregion showing the largest amount of degradation． This was characterised bythe development of a cathode－crack ratio （CCR） to quantify the extent of crack growthin the regions． Results showed that the CCR at 2000 and 5000 cycles was largestfor the outlet， followed by the middle region and the inlet CCR being the smallest．Further quantification of the crack formation was done by analysis of the crackconnectivity and label analysis． Results of these investigations highlighted furtherthe localised nature of the degradation， with the outlet and middle regions beingfound to have a greater connection of the crack network in the CL than the inlet．

To investigate the degradation at thenanoparticle level， TEM studies were done on a fresh electrode and comparedwith samples extracted from the three regions． The average particle diameter ofthe fresh electrode platinum particles was found to be 2．6 nm ± 0．4 nm， withaverage diameters found to increase for the degraded electrode， ranging between3．4 ± 0．4 nm at the inlet to 3．7 ± 0．5 nm in the middle， with the outletfalling between with a value of 3．5 nm ± 0．4．

Finally， analysis under the land andchannel of the three regions was found to show that the extent of crackingunder the land is less than under the channel， especially after 5000cycles， characterised by the increased crack width under the channel， as wellas a greater CCR under the channel is some cases．

The results shown here indicate that X－rayCT is a useful tool for quantifying the morphological degradation mechanismsalong－the－channel of a serpentine flow field and confirm emerging findings thatthere is non－uniform localised performance across the MEA， from inlet tooutlet． Furthermore， the results shown here indicate the need for intelligentMEA design， with graduated CL thickness or varying catalyst and carboncontent from inlet to outlet being two suggested methods for balancing thedegradation gradient across the MEA． Finally， investigations should bevalidated on larger operating cells， as well as including the quantification ofwater formation and probing a wider range of AST protocols， given that thedegradation of materials in the MEA is not limited to thecarbon－corrosion－specific degradation investigated in this work．