The objective was to examine the catalyst layer and the interaction between the membrane and catalyst layer of a polymer electrolyte membrane (PEM) fuel cell membrane electrode assembly (MEA).
The membrane itself is an ionomer-impregnated PTFE film that separates the anode and cathode sides of the fuel cell. The membrane is coated on both sides with a thin catalyst layer that consists of microscale carbon particles each supporting nanoscale platinum catalyst particles all loosely embedded in a matrix of ionomer. This catalyst-coated membrane is the MEA. The ionomer microstructure and ionomer-catalyst layer interface are important factors in the performance of the fuel cell. The ionomer microstructure determines the ion exchange across the membrane that allows the fuel cell reaction to occur, and the Pt particles that catalyze the fuel cell reaction must be properly distributed in the catalyst to maximize reaction efficiency.
The specimen examined was a generic MEA similar to that used in commercial fuel cell research and development:
Membrane: Dupont NafionTM 112 (perfluorosulfonic acid ionomer on a PTFE backbone)
Catalyst: both anode and cathode consisted of a dispersion of ~100 nm C particles supporting <5 nm Pt particles. Anode Pt loading was ~0.4 mg/cm2, cathode Pt loading was ~0.5 mg/cm2
Figure 1 - Photograph of a section of MEA
1. Critical Point Drying (CPD)
The membrane is strongly hydrophilic. Though it had never been exposed to liquid water, there was the concern that it may have absorbed atmospheric water vapor and could be damaged by the vacuum in the SEM. To circumvent this possibility, an attempt was made to use critical point drying (water » ethanol » CO2) to remove any absorbed water from the MEA. However, the catalyst layer is apparently unstable in ethanol (and other organic solvents). The drying attempt resulted in a mesh sample holder containing a clean membrane surrounded by a sea of very dry black powder. This was less than ideal for imaging, so CPD was not used in the imaged specimens.
The ionomer is nonconducting (for electrons) and requires a conformal, conductive coating for SEM imaging; thus, all SEM specimens were sputtercoated with ~40 Å of gold to prevent the beam from charging the sample.
All catalyst-ionomer interface sample cross-sections used were obtained by standard glass-knife microtomy. Due to the low energy (30 kV) of the beam used with the SEM transmission electron detector, samples would ideally be 50-75 nm thick. Unfortunately, the samples proved too fragile and tended to separate and fragment both macroscopically and microscopically when sectioned <100 nm. This shown in the figure below. Final samples used were 100-200 nm thick.
Figure 2 - Microtomy Damage to MEA
Left - 100x image of the specimen on a TEM grid showing macroscopic damage from microtomy
Right - 20,000x image of the same TEM specimen showing microscopic separation from microtomy.
The bright areas are due to gamma correction to highlight the separation and not due to sample charging.
Examination of the Surface of the Catalyst Layer
To verify that the MEA sample used was "standard" and compared well to other specimens, SEM images of the surface of the MEA were obtained and compared to images in the literature. This provided a useful check that no gross discrepancies existed in the fabrication and preparation of the specimen. A sample image and a comparison image from the literature are shown below:
Figure 3 - Comparison of Specimen Image to Literature Image
Left - 3000x image of the catalyst layer surface of the specimen MEA
Right - 3000x image of the catalyst layer surface of a "standard" MEA used in university research1
Note that the images are very similar with the notable exception of the flattened appearance of the specimen MEA. This is likely due to the process by which the catalyst layer is mounted on the membrane, which requires a decal to be hot-pressed to the membrane.
After verifying that the specimen was comparable to others, a more intensive surface examination was conducted. This examination imaged the catalyst layer at increasingly higher magnifications to capture the C particles. Since these particles are ~100 nm and are not well-defined, high magnification was necessary for imaging. The Pt particles themselves are too small (~5 nm) and too poorly defined to image well in the SEM.
Figure 4 - Imaging of C Particles in Catalyst Layer
Left - Catalyst layer surface at 20,000x; the myriad small spheres are C particles
Right - Same specimen at 100,000x; a few C particles have been highlighted
To allow the reader to get a general feel for the microstructure of the catalyst surface, a short movie was made of twenty-one images taken of the same area at successively increasing magnification (10x - 300,000x).
Examination of the Catalyst-Ionomer Interface
The catalyst-ionomer interface on an unused MEA should be fairly continuous and not show signs of migration of the Pt from the catalyst into the ionomer. To verify this, images were obtained of this boundary using SEM, TEM and an SEM with a Transmitted Electron Detector (TED).
Figure 5 - Comparison of Catalyst-Ionomer Boundary Layer Images Using Different Imaging Techniques
Left - SEM SE Detector Image at 44,000x (color inverted)
All three images are comparable. The SE and TED images show the general shape of the C particles (black blobs)
The superior resolution of the TEM allowed direct imaging of the Pt nanoparticles. High magnification examination of the interface found no sign of large-scale Pt migration into the membrane, as seen in the figure below.
Figure 6 - TEM Image of Catalyst Layer-Ionomer Interface at 500,000x
Though it isn't immediately apparent in this scanned image whether or not Pt migration is occurring,
it was easily seen during the TEM scan that there are no signs that any migration is taking place.
One of the goals of the TEM examination was to characterize the Pt particle size, shape, and distribution. Though it proved possible to image the Pt particles, it was difficult to get the resolution necessary to generalize their size and shape. Many of the particles in Fig 6 are spherical, but enough shapes are overlapped that it is impossible to tell if they are individual particles or a larger, misshapen particle (perhaps caused by sintering of the individual spheres). The Pt distribution is easier to see. In images such as Fig 5-right, it is readily seen that the Pt is fairly evenly distributed on each C support particle and very little Pt has dislodged from its support and become interstitial.
The electron beam should exhibit Bragg diffraction from the Pt particles in the catalyst layer. However, the C particles are amorphous and have no regular lattice, so it was expected that a diffraction pattern might be overwhelmed by the haze of the C particles. The truth turned out to be somewhere in the middle, where the haze overwhelms some rings and some are still visible.
Figure 7 - TEM Diffraction Pattern of Catalyst Layer
The image was severly overcontrasted in post-processing to show diffraction rings; the rings were nearly invisible on the original film negative.
The distance from the 000 beam to the first visible ring is measured at 9 mm at a camera length of 0.8 m and beam energy of 200 kV. From this, interplanar spacing d for this ring was calculated to be 2.23 Å. The theoretical value of d for Pt(111) is 2.26 Å, so this ring seems to be a good indicator of the Pt. However, the C haze overwhelms and merges other rings, making it difficult to measure the ring distance with any kind of reliability. Unfortunately, electron diffraction is limited in its usefulness under these conditions.
Though PEM fuel cell MEAs presents some sample preparation and imaging difficulties, electron microscopy provides a useful and necessary tool for basic MEA research.
The fragile nature of the MEAs in microtomy makes 200kV TEM much more useful than 30kV TED for nondestructive examination. The higher energy beam in TEM allows for a thicker (>100 nm) sample than is necessary for the TED. Higher end microtomy techniques might mitigate this difficulty.
The fuel cell models used to predict MEA behavior and advance the basic understanding of the catalyst and ionomer performance require validation and verification after MEA assembly, and SEM/TEM is well suited for verifying that the catalyst (Pt) and catalyst support (C) composition and distribution are as predicted.
Morikawa, Tsuihiji, Mitsui, Kanamura: Preparation of Membrane Electrode Assembly for Fuel Cell by Using Electrophoretic Deposition Process, Journal of The Electrochemical Society, 151 (10) A1733-A1737 (2004)
K.L. More: “Microstructural Characterization of PEM Fuel Cell Membrane Electrode Assemblies”, from the DOE Hydrogen Program FY 2004 Report section IV.I.2
Antoine & Durand: In Situ Electrochemical Deposition of Pt Nanoparticles on Carbon and Inside Nafion, Electrochemical and Solid-State Letters, 4(5) A55-A58 (2001)
Special thanks to Brian McIntyre of UR Optics for TEM imaging support and to General Motors Fuel Cell Activities for MEA samples and for use of the SEM w/ TED.