Failure analysis of fuel cell electrodes using three－dimensional multi－length scale X－ray computed tomography

A．Pokhrel
M．El Hannach
F．P．Orfino
M．Dutta
E．Kjeang

Abstract

X射线断层扫描技术用于燃料电池失效电极的三维微米纳米分析

X－ray computed tomography （XCT）， a non－destructive technique， is proposed for three－dimensional， multi－length scale characterization of complex failure modes in fuel cell electrodes． Comparative tomography data sets are acquired for a conditioned beginning of life （BOL） and a degraded end of life （EOL） membrane electrode assembly subjected to cathode degradation by voltage cycling． Micro length scale analysis shows a five－fold increase in crack size and 57％ thickness reduction in the EOL cathode catalyst layer， indicating widespread action of carbon corrosion． Complementary nano length scale analysis shows a significant reduction in porosity， increased pore size， and dramatically reduced effective diffusivity within the remaining porous structure of the catalyst layer at EOL． Collapsing of the structure is evident from the combination of thinning and reduced porosity， as uniquely determined by the multi－length scale approach． Additionally， a novel image processing based technique developed for nano scale segregation of pore， ionomer， and Pt／C dominated voxels shows an increase in ionomer volume fraction， Pt／C agglomerates， and severe carbon corrosion at the catalyst layer／membrane interface at EOL． In summary， XCT based multi－length scale analysis enables detailed information needed for comprehensive understanding of the complex failure modes observed in fuel cell electrodes．

电压循环造成的电压衰减

微米尺度上，阴极催化层的连续性变差，裂缝宽度增加。

统计的结果，和图像的宏观结果一致。

需要进一步分析的样品块，从宏观到微观。

纳米尺度上，阴极催化层的状态，孔变少，孔变大。厚度由9微米变为4．5微米。

The proposed three－dimensional， non－destructive multi－length
scale visualization approach utilized a ZEISS Xradia 520 Versa
XCT system （Versa） at the micro length scale and a ZEISS Xradia 810
Ultra XCT system （Ultra） at the nano length scale．

微米尺度和纳米尺度使用了不同的X射线断层扫描设备。

微米尺度扫描使用9小时。

纳米尺度扫描使用52小时。

图像处理分析获得的孔容。

根据孔结构仿真获得的有效扩散系数，而且呈现各项异性的现象，TP方向的扩散系数下降更加显著。

数据图像处理获得的孔，树脂和催化剂的占比

催化剂黄色，孔黑色，树脂蓝色，表明三维尺度上材料出现相分离。靠近膜的一侧的催化层树脂含量增加，催化剂向气体扩散层一侧聚集长大。

Conclusions

3D failure analysis of fuel cell electrodes was performed by using
X－ray computed tomography． A multi－length scale approach was
used to investigate crack growth， thinning， and structural change in
the catalyst layer． This was accomplished by analyzing an end of life
MEA subjected to 4700 potential cycles from 0．6 V to 1．3 V
compared to a pristine， conditioned MEA． The micro length scale
analysis showed 57％ thinning of the cathode catalyst layer
accompanied by a five－fold increase in crack size， indicating
widespread action of carbon corrosion． It was concluded that
catalyst layer crack growth and thinning can occur in parallel with
comparable dimensional rates， which can be attributed to a combination of carbon corrosion and erosion from water flooding．

这里对于水淹使用的是物理的磨蚀。

The complementary nano length scale analysis of the cathode
catalyst layer showed a reduction in porosity from 54％ at BOL to
23％ at EOL with comparatively more isolated and larger pores in
the degraded CCL． The combination of microscopic reduction in
porosity and macroscopic reduction in thickness， uniquely
observed through the present multi－length scale failure analysis
approach， conclusively verifies collapsing of the catalyst layer
structure as a result of severe carbon corrosion and associated loss
of physical integrity． Diffusion simulations inside the reconstructed
nano scale structure showed dramatic reduction of the effective
diffusivity of oxygen and water vapor in the degraded material．
Thus， in terms of diffusive mass transport to and from the active
sites， the reduced porosity and more isolated pores at EOL
considerably outweighed the positive effect of the larger pore size．
In contrast to BOL， the EOL effective diffusivities were found to be
anisotropic in the three principal directions， with the through plane
diffusivities being substantially lower than in the in plane direction．
The low EOL fuel cell performance at high current densities can
thus be attributed to mass transport losses in the degraded cathode．

     In addition， a novel image processing based technique was
developed for three－phase segmentation of the 3D reconstructed
nano scale structure into pore， Pt／C， and ionomer dominated phases．
The ionomer content was assumed to be retained during
voltage cycling while the pore and Pt／C dominated phases were
reduced in volume by collapsing and dissolution／corrosion，
respectively． The ionomer dominated phase was observed to morph
from a thin film structure at BOL to a dense structure with large，
interconnected features of several hundred nanometers in diameter
that dominate the remaining solid phase at EOL． The dense
ionomer structure reduces the space available for reactant transport
in the pore phase and electron conduction in the Pt／C phase
and is thus expected to contribute to both mass transport and
ohmic losses， as observed from the fuel cell performance at EOL．
Particle isolation and agglomeration was observed in the remaining
Pt／C dominated phase in the degraded structure， which is in good
qualitative agreement with the high ECSA loss measured by cyclic
voltammetry and the increased Pt particle size measured by XRD，
and can be attributed to extensive loss of carbon support combined
with catalyst dissolution via Ostwald ripening and／or coalescence
of Pt nanoparticles． Furthermore， carbon corrosion and Pt dissolution were observed to be comparatively more severe at the membrane side of the catalyst layer， which further contributes to
the high kinetic performance losses measured at EOL．
    Overall， the 3D multi－length scale approach demonstrated in
this work provided detailed and valuable insight into the failure
mode of the cathode catalyst layer via micro scale analysis of crack
growth and thinning combined with nano scale analysis of structural
and compositional features that cannot be observed through
conventional diagnostic techniques． The new insight regarding the
complex electrode structure enabled by this approach is expected
to be useful for a wide range of failure analysis assignments from
lab－scale research to field operation of fuel cells， and can also be
extended to other electrochemical devices such as batteries and
supercapacitors．