83

Chapter 4 Integrated Pt/CNT-based Electrocatalyst for PEMFCs

4.1 Introduction
Following our optimization studies on the in situ growth of CNTs on carbon
paper shown in Chapter 3, the application of these in situ grown CNTs is presented in
this chapter where fabrication and characterization of an integrated Pt/CNT-based
electrocatalyst are intensively introduced. The integrated Pt/CNT-based
electrocatalyst was fabricated for PEMFC electrodes via direct Pt sputter-deposition
and their electrocatalytic performance on PEMFC reactions was characterized in a
real PEMFC system. The aim of this work was to optimize the fabrication method for
the integrated Pt/CNT-based electrode as well as to evaluate the effectiveness of this
electrode as an integrated PEMFC component for high efficiency PEMFCs.

It has been reported in a large number of studies that Pt/CNT-based
electrocatalysts demonstrated higher electrochemical activity and stability than those
of the conventional Pt/VXC72R electrocatalyst [1-8]. To exploit such improvement of
Pt/CNT-based electrocatalysts, several research groups developed their Pt/CNT-based
electrocatalysts based on in situ grown CNTs on carbon paper as the catalyst support
[9-12]. The synthesis processes of the in situ grown CNTs in their studies were
described in Section 3.1 and this section mainly introduces their synthesis methods for
Pt deposition onto the in situ grown CNTs. In earlier study by Wang and coworkers
[9], Pt catalysts were electrodeposited onto the in situ grown multi-walled carbon
nanotubes (MWNTs) by a three-electrode DC method in 5 mM H
2
PtCl
6
and 0.5 M
H
2

SO
4
aqueous solution. The deposition process was carried out at a potential of 0 V
vs. SCE and the Pt loading was controlled by the total charge applied. It was reported
84

that a total Pt loading of 0.2 mg cm
-2
was obtained by this electrodeposition method
and the average Pt particle size was around 25 nm. However, the large Pt particles
were found to be the major handicaps that caused a lower polarization performance
compared to that of the conventional ink-process prepared electrode, where the Pt
particle size was in the range of 2–4 nm. To reduce the Pt particle size, they used a
chemical reduction method in their subsequent work to synthesis Pt/CNT-based
electrocatalysts [10]. Prior to Pt deposition, the in situ grown CNTs were first surface
functionalized by refluxing the CNT-grown carbon paper in 4 N H
2
SO
4
/4 N HNO
3

for 3 h. To deposit Pt catalysts, a certain amount of Pt precursor solution was sprayed
onto the CNT-grown carbon paper under a 60−80 °C heating condition, and then
followed by reduction in 20% H
2
in N
2
at 150 °C for 2 h. The Pt precursor solution
was a mixture of 250 mg 8 wt% H

2
PtCl
6
, 60.8 mg 5 wt% Nafion solutions and 4 g
isopropanol. After Pt deposition, approximately 0.15 mg cm
-2
Pt was deposited onto
the in situ CNTs as determined the weight difference of the CNT-grown carbon paper.
According to the TEM micrograph of the deposited Pt catalysts, the Pt particle size
was reduced to 4 nm via this chemical reduction method and enhanced Pt utilization
was attained to give higher polarization performance. However, it was found that the
polarization performance of the Pt/CNT-based electrode remained very low without
brushing an additional gas diffusion layer on the backside of the carbon paper.

Based on Wang’s exploration on Pt catalysts supported on in situ grown CNTs,
in 2006 Villers et al. also prepared a Pt/CNT-based electrocatalyst for PEMFC
electrodes [11]. In their method, the in situ grown MWNTs were first immersed in a
1% silane solution diluted in 93% ethanol and 6% H
2
O, which consisted of 0.04 M
PtCl
4
and 0.04 M PtBr
4
. After 2 h immersion, the MWNT-grown carbon paper with Pt
85

precursor were then air-dried and chemically reduced in H
2
at 500 °C for 15 min.

After reduction, they found that the deposited Pt catalysts were nanosized particles
with an average particle size of 3−5 nm. However, chemical state analysis of the Pt
nanoparticles by XPS revealed that there were still some Pt(II) (22%) and Pt(IV) (4%)
present after Pt deposition. Moreover, it was found that one deposition process
corresponded to a Pt loading of 0.16 mg cm
-2
and increasing Pt loading was achieved
by repetitive deposition, which indicated that this Pt deposition process could not
provide efficient control of Pt catalyst loading. Later Saha and coworkers [12]
claimed that a high Pt loading on in situ grown CNTs could be obtained using glacial
acetic acid as the reducing agent. In their study, the MWNT-grown carbon paper was
first chemically oxidized in a 5 M HNO
3
aqueous solution for 5 h before Pt deposition.
Afterward, the MWNT-grown carbon paper was washed with deionized water and
dried in vacuum at 90 °C for another 5 h. During the Pt deposition process, the
MWNT-grown carbon paper was immersed into a mixture dispersion of Pt
acetylacetonate (Pt(acac)
2
) and 25 ml glacial acetic acid and ultrasonicated for 2 min
at room temperature. Then it was heated up to 110 °C for Pt reduction and held for 5 h
with constant stirring. At last, the Pt-deposited MWNT-grown carbon paper was
rinsed with deionized water and dried at 90 °C overnight in a vacuum oven. The Pt
loading obtained via this process was around 0.42 mg cm
-2
determined by inductively
coupled plasma-optical emission spectroscopy (ICP-OES). Although the deposited Pt
nanoparticles showed high density and small size distribution range (2−4 nm) on the
in situ MWNTs, it should be noted that this deposition process was rather tedious and
time-consuming. In addition, the wet-chemical deposition methods described above

all revealed difficulties in control of Pt loading and Pt distribution, considering that
simultaneous Pt deposition may occur on the carbon fibers in carbon paper, making
86

these Pt catalysts inaccessible to reactant gases. Furthermore, despite the enhanced
polarization performance observed from the Pt/CNT-based electrocatalysts prepared
by the above wet-chemical methods, it is worth noting that an additional VXC72R-
based gas diffusion layer was always needed on the backside of the carbon paper that
adds further complexity to electrode preparation. In view of all these limitations, the
effectiveness of the in situ grown CNTs as catalyst support is greatly undermined by
the Pt deposition process.

To resolve the above mentioned difficulties in Pt deposition onto in situ grown
CNTs, controllable Pt deposition was conducted via direct sputter-deposition in this
study. Sputter-deposition technique has the advantages of being able to directly
deposit Pt catalyst with excellent control of deposition rate, as demonstrated in a
number of studies [4, 13-17]. Moreover, as it is a surface deposition technique, it
allows us to disperse the Pt catalyst highly localized at the surface of the CNT-grown
carbon paper, leading to higher Pt utilization [4]. Last but not least, sputter-deposition
technique was chosen for Pt deposition owing to its ability to directly deposit Pt
catalysts onto the dense in situ grown CNT layer, thus additional CNT surface
oxidation process can be eliminated. In the following sections, the fabrication process
of the Pt/CNT-based electrocatalyst via sputter-deposition will be introduced and the
overall effectiveness of this Pt/CNT-based electrocatalyst will be evaluated by a series
of ex situ and in situ characterization studies. This chapter mainly concentrates on the
electrochemical activity and mass transport properties of the Pt/CNT-based
electrocatalyst for PEMFC electrodes. The electrochemical stability of this Pt/CNT-
based electrocatalyst will be demonstrated and elaborated in the subsequent chapter.
87


4.2 Fabrication and Characterization of Integrated Pt/CNT-based
Electrocatalyst
This section presents the fabrication process of the sputter-deposited Pt catalysts,
as well as the structural and compositional characterization towards them. The
structural and compositional analysis is of much importance in evaluating the
integrated Pt/CNT-based electrocatalyst in that it can provide valuable insights with
regard to understanding the electrocatalytic performance of the catalyst in addition to
the electrochemical characterization.

4.2.1 Fabrication of Integrated Pt/CNT-based Electrocatalyst
In this study, direct sputter-deposition of Pt catalysts onto the in situ grown
CNTs was carried out using a R.F. magnetron sputtering system (Denton Discovery-
18). In contrast to previous studies where wet-chemical deposition processes were
used [9-12], several pieces of CNT-grown carbon papers were directly transferred into
the sputtering chamber after CNT growth without surface oxidation. During the
sputtering process, the Ar gas pressure was maintained at 10 mTorr while the specific
deposition rate of Pt catalysts was varied by controlling the output sputter-power. The
specific deposition rate of the Pt catalysts at a given output power was determined by
weight difference of the CNT-grown carbon paper before and after Pt deposition.
Controllable Pt deposition was then realized by varying the sputtering time based the
calculated specific deposition rate. Initially a Pt loading of 0.04 mg cm
-2
obtained at
100 W output sputter-power was used in this study, which is one tenth of the Pt
loading in a commercial Pt/VXC72R-based electrode [18].


88

4.2.2 Characterization of Integrated Pt/CNT-based Electrocatalyst

In order to reveal the microstructure and morphology of the sputter-deposited Pt
catalysts on the in situ grown CNTs, their TEM micrographs were examined using a
high-resolution JEM-2010 FETEM system. Figure 4.1 (a) shows the TEM micrograph
of the CNT grown on carbon paper at a C
2
H
4
flow rate of 20 sccm. It can be seen that
the CNTs grown on carbon paper by the thermal CVD process were multi-walled
carbon nanotubes (MWNTs) with a typical outer diameter of about 30 nm and an
inner diameter of around 10 nm. The Fe catalysts can also be observed as small
particles buried inside the tube, which agrees well with previous studies [19, 20].
Moreover, it is noteworthy that the in situ grown MWNTs showed very coarse and
incontinuous external graphitic walls, suggesting that various defects were generated
on the CNT skin during growth. This result is in excellent agreement with the Raman
spectra study of the in situ grown CNTs shown in Section 3.3.2. Nevertheless, the
defects on the CNT surface may be a favorable feature for Pt deposition due to the
presence of numerous anchoring sites for Pt particles, in contrast to the inertness of a
perfect CNT skin that was reported unable to be wetted by liquids with a surface
tension higher than 100–200 mN m
-1
[21]. The in situ grown MWNT layer was then
coated with 0.04 mg cm
-2
Pt catalysts by sputter-deposition without any surface
oxidation. Figure 4.1 (b) and (c) illustrate the TEM micrographs of the as-deposited Pt
catalysts on in situ grown MWNTs. Contrary to previous studies where Pt catalysts
prepared by various chemical reduction methods usually showed poor dispersion on
untreated CNTs [10, 21], numerous nanosized Pt particles were observed to be
densely dispersed on the CNT support via direct sputter-deposition as shown in Fig

4.1 (a). Instead of forming a Pt thin film, as was seen on the smooth Si substrate, the
sputter-deposited Pt catalyst showed a scaly structure with a homogeneous
89

distribution on the CNT layer due to its high surface roughness and porosity. The
average particle size of the Pt nanoparticles was approximately 2−3 nm. The Pt
loading of 0.04 mg cm
-2
corresponds to a Pt thin film of thickness about 20 nm on a
smooth Si substrate. Given the extremely high roughness and porosity of the in situ
grown MWNT layer, it is understandable that the Pt layer on MWNTs was of much
lower thickness and thus Pt nanoparticles were formed and uniformly dispersed on the
MWNT surface. According to Fig. 4.1 (c), the grain size distribution of the Pt
nanoparticles on a single CNT shows a relatively small range mostly from 1−5 nm.
TEM investigation demonstrates that well-dispersed Pt nanoparticles have been
successfully deposited onto the in situ grown CNTs by direct sputter-deposition,
which provides significant advances over the wet-chemical processes that Pt particle
distribution is greatly enhanced and CNT surface oxidation is eliminated.















(a)
(b)
90

0123456
0
10
20
30
40
50
60


Frequency / %
Diameter / nm









Figure 4.2 shows the corresponding size distribution histogram based on Fig. 4.1
(c). The total number of the counted Pt nanoparticles was around 500−550 to ensure a
statistically representative sampling. As observed in Fig. 4.2, most of the sputter-

deposited Pt nanoparticles were less than 4 nm in diameter, with the majority
distribution in 2−3 nm. It was reported previously by Giordano et al. [22] that the









Fig. 4.1 TEM micrographs of the in situ grown CNTs (a) before and (b) after
Pt sputter-deposition, (c) Pt nanoparticles on a CNT support. Pt loading: 0.04
m
g
c
m
-2
(c)
Fig. 4.2 Histogram of Pt particle size distribution based on Fig. 4.1 (c)
91

mass activity of Pt catalyst has a strong correlation to the Pt particle size and the
maximum mass activity corresponds to a grain size around 3 nm. Therefore the
sputter-deposited Pt catalysts may give rise to a high mass activity for PEMFC
reactions based on the particle size distribution demonstrated in Fig. 4.2.























Pt polycrystalline
MWNT
Graphite
layers
(b)
(a)
Fig. 4.3 HRTEM micrographs of (a) an as-grown CNT tip and
(b) a CNT tip with sputter-deposited Pt catalysts.
92

To further investigate the microstructure of the Pt/CNT-based electrocatalyst,
high-resolution TEM (HRTEM) micrographs were obtained at magnifications up to

500,000. Figure 4.3 (a) shows the multi-graphitic walls on the tip of a CNT that was
grown on carbon paper. The layered graphitic walls were clearly seen at the closed
end with Fe catalysts buried inside, indicating a tip-growth mechanism for the in situ
grown CNTs [19]. Considerable defects were also observed at the CNT surface,
corresponding to the Raman results shown in the precious chapter. After Pt sputter-
deposition, the TEM micrograph of a CNT tip with sputter-deposited Pt nanoparticles
is illustrated in Fig. 4.3 (b). It is noticeable that the Pt nanoparticles exhibited a clear
lattice fringe of crystallite structure, suggesting that the Pt catalyst produced via
sputtering technique has a high crystallinity. This implies that the sputter-deposited Pt
catalysts are probably in a pure metallic state with little oxide content.

To examine the chemical state of the sputter-deposited Pt catalysts, XPS
analysis was performed on the Pt/CNT-based electrocatalyst as shown in Fig. 4.4.
Determination of Pt chemical state was carried out by means of spectrum
deconvolution using a Gaussian/Lorentzian shape line modified by an asymmetric
function. As can be seen in Fig. 4.4, the Pt 4f
7/2
core level revealed a binding energy
of 71.1 eV based on the Pt/CNT composite catalyst, with reference to 284.4 eV as the
binding energy of C 1s. This result substantially supports the TEM results that the
sputter-deposited Pt catalyst is mostly in the pure metallic state and has a fine
crystallite structure. However, it is likely that a minute amount of oxidized Pt is
present due to the anchoring oxidized groups on the CNT surface. By contrast, Villers
found a noticeable amount of oxidized Pt catalysts on the in situ grown CNTs that
were deposited by wet-chemical methods [11]. Therefore, a reduction process for the
93

84 82 80 78 76 74 72 70 68 66 64
0
4000

8000
12000
16000
20000


CPS
Binding energy / eV
4f
7/2
, 71.1 eV
4f
5/2
, 74.4 eV
post-deposited Pt particles was indispensable in their study. The vast composition
difference of the Pt catalysts may probably be due to the intrinsic properties of the
deposition methods. In a typical chemical reduction process, the content of Pt oxides
in the synthesized Pt catalyst greatly depends on the reducing agent used for the Pt-
based solvent solution [11, 12, 23]. Pt oxides are likely to form during deposition
when the reduction of Pt precursor is insufficient. While in the sputtering process, Pt
atoms are ejected from a pure Pt target and directly deposited onto CNT/carbon paper
substrates in a high-vacuum environment. This process particularly reduces the
possibilities of Pt oxide formation thus showing a great potential for Pt deposition for
PEMFC applications as demonstrated in previous studies [13-17].














Fig. 4.4 XPS spectrum of Pt nanoparticles on CNT/carbon paper by
sputter-deposition. X-ray source: Al Kα 1486.6 eV, pass energy: 20 eV.
94

4.3 Integrated Pt/CNT-based Cathode for PEMFCs
As described previously in Chapter 1, the oxygen reduction reaction (ORR)
occurring at PEMFC cathode is the rate determining step for PEMFC operation.
Under normal conditions the ORR kinetics is very slow thus Pt-based catalysts are
commonly used at the electrocatalyst for this reaction. However, the ORR kinetics is
still rather sluggish compared with the hydrogen oxidation reaction (HOR) that most
of the activation overpotential derives from the ORR while the contribution of the
HOR is usually negligible [24]. As such, in this study we focus on the integrated
Pt/CNT-based electrocatalyst for the ORR by examining their in situ electrochemical
performance as the PEMFC cathode electrocatalyst.

4.3.1 Optimization of Electrode Preparation
To actively verify the electrochemical performance of the integrated Pt/CNT-
based for the ORR, electrode preparation process was first optimized in terms of a
series of experimental parameters, including Nafion impregnation, sputtering power,
Pt catalyst loading and CNT layer morphology. In a typical MEA used in this study
the integrated Pt/CNT-based electrocatalyst was made into the cathode while the
anode was a conventional VXC72R-based gas diffusion electrode with a commercial
Hispec4000 catalyst (40 wt% Pt/VXC72R, Johnson-Matthey). The Pt loading at

anode was maintained at 0.2 mg cm
-2
and the electrode preparation process for the
conventional VXC72R-based electrode was described in Section 2.2.3. For the
Pt/CNT-based cathode, 1 mg PTFE was brushed onto the back side of the
CNT/carbon paper backing before Pt deposition, in order to provide electrode
hydrophobicity. The optimization of electrode preparation process for the Pt/CNT-
95

based cathode were carried out mainly based on polarization curve characterization,
of which results are shown below in sequence.

Nafion Impregnation
It is well-known that Nafion impregnation into catalyst layer can significantly
improve cell performance in that the three-phase reaction zone is greatly increased by
impregnating Pt catalysts with Nafion electrolyte. Paganin et al. [25] investigated the
effect of Nafion impregnation on commercial Pt/VXC72R-based electrode and found
that the cell performance notably improved when the Nafion loading was increased
from 0.87 to 2.2 mg cm
-2
. They also found that the cell performance reached a
maximum when the Nafion loading was equivalent to 33 wt% of the catalyst layer
weight. Later Qi et al. [26] also observed this optimum Nafion loading that the weight
ratio between Nafion and Pt/VXC72R was 1:2 for conventional ink-process prepared
electrodes. To find out the optimum Nafion loading for the Pt/CNT-based cathode in
this study, the effect of Nafion impregnation was investigated by spraying 0.5 wt%
Nafion solutions onto several Pt/CNT-based electrodes bearing different Nafion
loadings. The corresponding polarization curves are shown in Fig. 4.5 below.

As can be seen in Fig. 4.5, the cell performance was visibly enhanced when 40

µl 0.5 wt% Nafion solutions were sprayed onto the Pt/CNT-based cathode, compared
to that of the MEA without Nafion impregnation. With increasing Nafion
impregnation, a maximum cell performance was achieved from the MEA with a total
amount of 60 µl Nafion solution sprayed at cathode, corresponding to the optimum
Nafion loading of 0.05 mg cm
-2
. This MEA could give rise to a maximum power
density of 610 mW cm
-2
based on 0.04 mg cm
-2
Pt at cathode, owing to the increased
96

proton conductivity within the catalyst layer. However, when the Nafion loading
further increased, it was observed that the cell performance dropped dramatically at
large current densities. It can be speculated that the voltage drop at large current
density region may be probably due to the wrapping of Pt catalysts by the excess
Nafion ionomers, causing increased mass transport resistance.














Sputter Output Power
The effect of sputter output power was also investigated as a parameter for the
optimization of electrode preparation. Initially the sputter-deposition of Pt catalysts
was carried out at an output power of 100 W, which gives a specific Pt deposition rate
of 35 µg cm
-2
min
-1
. In this study, a series of sputter output powers, including 50, 30
and 15 W, were used to deposit Pt catalysts and their effects on cell performance were
evaluated based on their polarization curves. The impregnated Nafion loading was 60
0.0 0.5 1.0 1.5 2.0
0.0
0.2
0.4
0.6
0.8
1.0


Cell potential / V
Current density / A cm
-2
No Nafion impregnated
40ul 0.5wt% Nafion impregnated
60ul 0.5wt% Nafion impregnated
80ul 0.5wt% Nafion impregnated
120ul 0.5wt% Nafion impregnated

Pt loading: 0.04 mg cm

at cathode

Fig. 4.5 Polarization curves of Pt/CNT-based cathodes
with various Nafion loadings.
97

µl of 0.5 wt% Nafion solution for all the samples. As shown in Fig. 4.6, the Pt/CNT-
based electrodes obtained at 50 and 30 W sputter powers demonstrated similar cell
performance, which was noticeably higher than those of the electrodes prepared under
100 and 15 W sputter powers. According to Huang et al. [27], lowering sputter power
could produce more refined Pt particles at lower sputtering rate. However, they found












that lower sputter power also reduced the kinetic energy of the ejected Pt atoms,
leading to a diminished penetration depth of the Pt particles. Accordingly, it is very
likely that the sputter-deposited Pt catalysts at 50 and 30 W had smaller grain sizes
than at 100 W, as well as greater penetration depths into the CNT layer than at 15 W,
owing to the higher kinetic energy of the sputtered Pt atoms. The specific Pt

deposition rates for 50 and 30 W sputter powers were found to be 10 and 5 µg cm
-2

min
-1
, respectively. Therefore 50 W was chosen as the optimum sputter power for the
0.0 0.5 1.0 1.5 2.0 2.5
0.0
0.2
0.4
0.6
0.8
1.0


Cell potential / V
Current density / A cm
-2
100W sputter output power
50W sputter output power
30W sputter output power
15W sputter output power
Pt loading: 0.04 mg cm

at cathode

Fig. 4.6 Polarization curves of Pt/CNT-based electrodes with sputter-
deposited Pt catalysts prepared at different output powers.
98


subsequent Pt deposition, whereby a typical Pt loading of 0.04 mg cm
-2
can be
obtained from a 4 min sputter-deposition process.

Pt Catalyst Loading
A number of studies have shown that sputter-deposition technique has great
potential for fabricating ultra-low Pt loading electrodes for PEMFC applications [13-
17]. The key advantage of this technique lies in that it is able to deposit Pt
nanostructures directly onto the rough electrode surface, where the reactant gases,
catalysts, and electrolyte form the three-phase zone. Hence the Pt utilization is
considerably enhanced by sputter-depositing Pt catalysts into the electrode-electrolyte
interface with an ultra-low Pt loading. Previously in this study the Pt/CNT-based
electrodes were all sputter-deposited with 0.04 mg cm
-2
Pt catalysts. In order to probe
the cell performance of Pt/CNT-based electrodes with different Pt loadings, a set of
electrodes were investigated on their polarization curves, where the Pt loadings were
0.02, 0.03, 0.04 and 0.06 mg cm
-2
, respectively. To prepare the Pt/CNT-based
electrodes, carbon papers with in situ CNTs grown at 20 sccm C
2
H
4
were used as the
sputtering substrates. The sputter output power was 50 W and the Ar gas pressure was
10 mTorr during Pt sputter-deposition. Prior to MEA assembly, the Pt/CNT-based
electrodes were brushed with 1 mg PTFE at the backside and air-sprayed with 60 µl
of 0.5 wt% Nafion solution on the electrode surface.


Figure 4.7 shows the polarization curves of the Pt/CNT-based electrodes with
various Pt loadings at cathode. It can be observed that the cell performance was
improved with increasing Pt loading at cathode. This is expectable that a larger
amount of Pt catalysts provides more reaction sites for the ORR. However, it was
99

noted that only slight improvement in cell performance was obtained when the
cathode Pt loading was increased from 0.04 to 0.06 mg cm
-2
. The maximum power
densities from 0.02, 0.03, 0.04 and 0.06 mg cm
-2
Pt catalysts were found to be 605,
635, 670 and 680 mW cm
-2
, respectively.












TEM micrographs of the in situ grown CNTs with different Pt loadings are

shown in Fig. 4.8 to reveal the microstructure of the Pt/CNT-based catalysts. It can be
clearly seen that the particle size and density of the sputter-deposited Pt catalysts grew
with increasing Pt loadings from 0.02 to 0.06 mg cm
-2
. When 0.02 mg cm
-2
Pt
catalysts were sputter-deposited onto the in situ grown CNTs as shown in Fig. 4.8 (a),
small Pt nanoparticles were observed on the CNT surface exhibiting a low distribution
density. As the Pt loading increased to 0.04 mg cm
-2
(see Fig. 4.8 (c)), it is noticeable
that the Pt nanoparticles formed numerous nanoscaled Pt islands and densely
dispersed on the CNT surface. As a result, the corresponding Pt surface area is greatly
Fig. 4.7 Polarization curves of Pt/CNT-based electrodes with
different Pt loadings at cathode.
0.0 0.5 1.0 1.5 2.0
0.0
0.2
0.4
0.6
0.8
1.0


Cell potential / V
Current density / A cm
-2
0.02 mg cm


Pt at cathode
0.03 mg cm

Pt at cathode
0.04 mg cm

Pt at cathode
0.06 mg cm

Pt at cathode

100




















enhanced by the increased Pt loading, leading to a visibly improved cell performance
as demonstrated in Fig. 4.7. However, it was revealed that the Pt islands coalesced
into large grains and fully covered the CNT surface as the Pt loading was further
increased to 0.06 mg cm
-2
. This result superbly explains the limited improvement in
cell performance from the Pt/CNT-based electrode with 0.06 mg cm
-2
Pt catalysts that
the Pt/CNT catalysts can only give slight increment in total Pt surface area when
Fig. 4.8 TEM micrographs of in situ grown CNTs with (a) 0.02,
(b) 0.03, (c) 0.04, and (d) 0.06 mg cm
-2
Pt catalysts.
(a)
(b)
(c)
(d)
101

higher Pt loadings are sputter-deposited beyond 0.04 mg cm
-2
. As such, the optimum
Pt loading was determined as 0.04 mg cm
-2
for our subsequent experiment.

CNT Layer Morphology
As previously depicted in Chapter 3, the in situ grown CNT layer at different

C
2
H
4
flow rate showed distinct surface morphologies on carbon paper. It allows
tunable surface roughness and porosity of the in situ grown CNT layers for Pt sputter-
deposition. The effect of CNT layer morphology has not yet been reported in previous
studies where CNTs were also grown on carbon paper to serve as Pt catalyst support
[9-12]. In this study, we investigated the cell performance of Pt/CNT-based electrodes
based on different in situ grown CNT layers with varied surface morphologies. The in
situ grown CNT layers on carbon paper were obtained from the optimized CNT
growth described in previous chapter at a series of C
2
H
4
flow rates of 5, 10, 15, 20
and 25 sccm, respectively. Their SEM images are illustrated in Fig. 3.8. The electrode
preparation parameters used, such as Nafion loading, Pt sputtering power and Pt
loading, were based on aforementioned optimization studies.

Figure 4.9 shows the polarization curves of Pt/CNT-based electrodes with Pt
catalysts sputter-deposited on different in situ grown CNT layers. As can be seen in
Fig. 4.9, a notable improvement in cell performance was achieved by using CNT
support layers grown at increasing C
2
H
4
flow rate from 5 to 20 sccm. However, this
tendency reversed as the C
2

H
4
flow rate was further increased to 25 sccm. According
to the surface morphology of the in situ grown CNT layers as revealed by their SEM
images, the dramatic difference in cell performance can be probably attributed to the
distinct surface morphologies of the CNT layers in view of their surface area and
102

porosity. It has been demonstrated that the surface area and porosity of the in situ
grown CNT layers were enhanced tremendously with increasing C
2
H
4
flow rate
during growth. Such enhancement would provide a considerably enlarged surface area
for the sputter-deposited Pt catalysts. However, it should be also noted that the pore
size of the CNT layers greatly decreased with increasing C
2
H
4
flow rate, due to which
the penetration depth of the sputtered Pt particles might be limited when the pore size
became extremely small. Consequently, it is likely that the Pt catalysts were sputter-
deposited mainly onto the top surface of the CNT layer, resulting in a reduced Pt
surface area for the ORR. This may probably be the reason for the cell performance
deterioration from the CNT support layer grown at 25 sccm C
2
H
4
flow. Accordingly

the in situ grown CNT layer obtained from 20 sccm C
2
H
4
flow can provide an
optimum surface morphology where both high surface area and large penetration
depth of Pt catalysts are guaranteed.












0.0 0.5 1.0 1.5 2.0
0.0
0.2
0.4
0.6
0.8
1.0


Cell potential / V
Current density / A cm

-2
Pt/CNT-based cathode (5 sccm C
2
H
4
)
Pt/CNT-based cathode (10 sccm C
2
H
4
)
Pt/CNT-based cathode (15 sccm C
2
H
4
)
Pt/CNT-based cathode (20 sccm C
2
H
4
)
Pt/CNT-based cathode (25 sccm C
2
H
4
)
Pt loading: 0.04 mg cm
-2
at cathode
Fig. 4.9 Polarization curves of Pt/CNT-based electrodes with

different in situ grown CNT layers as catalyst support.
103

4.3.2 Electrochemical Characterization of Integrated Pt/CNT-based Cathode
Subsequent to the optimization studies for electrode preparation, electrochemical
characterization of the integrated Pt/CNT-based cathode was carried out by means of
a series of in situ electrochemical tests, including polarization curve measurement, in
situ cyclic voltammetry (CV), and electrochemical impedance spectroscopy (EIS). To
evaluate the electrochemical performance of the Pt/CNT-based electrodes, two
reference electrodes were also prepared using commercial 20 wt% Pt on VXC72R (E-
TEK) and 40 wt% Pt on VXC72R (Johnson Matthey) catalysts. The details of
electrode preparation process have been described in Section 2.2.3. The three
electrodes with different catalysts were used as the cathode with an identical Pt
loading of 0.04 mg cm
-2
, while the anode was a conventional VXC72R-based
electrode for all of them with 0.2 mg cm
-2
Pt prepared by ink-spread method. Nafion
112 (Dupont Inc.) was used as the polymer electrolyte.

Figure 4.10 shows the polarization curves of the three electrodes with the
Pt/CNT catalyst and two commercial Pt/VXC72R catalysts, respectively. It can be
clearly seen that the Pt/CNT-based electrode showed a pronounced improvement in
the entire voltage region while the 40 wt% Pt/VXC72R catalyst gave the lowest
polarization performance among the three catalysts. This implies that the Pt/CNT-
based electrode may have enhanced charge transfer and mass transport properties
compared to those of the commercial Pt/VXC72R catalysts, thus leading to smaller
activation, ohmic and mass transport overpotentials as shown in Fig. 4.10. It was
found that the maximum power density based on the total Pt loading (0.24 mg cm

-2
)
was 2.8 W per mg of Pt for the Pt/CNT-based MEA, compared to 2.5 and 2.1 W per
mg of Pt for the two Pt/VXC72R-based MEAs. Furthermore, it should be noted that
104

0.00.51.01.52.02.5
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0

Power density / W cm
-2
Cell potential / V
Current density / A cm
-2
Pt/CNT catalyst at cathode
20 wt% Pt/VXC72R catalyst (E-TEK) at cathode
40 wt% Pt/VXC72R catalyst (Johnson Matthey) at cathode
Pt loading: A/C - 0.2/0.04 mg cm
-2


the polarization curves of the Pt/VXC72R-based electrodes showed gradual
deterioration usually after 10 curve scans, particularly at large current regions. This is
probably due to water flooding occurring at the cathode. By contrast, the Pt/CNT-
based electrode showed increasing polarization performance in the first 10 ~ 15 scans
derived from the activation of the sputter-deposited Pt catalysts, and thereafter its












polarization curves were rather stable within the test scope of 30 scans in this study. It
is likely that the inherent hydrophobicity and highly porous structure of the CNT layer
may provide superior water transport property of the electrode to prevent water
accumulating in the catalyst layer. Therefore, the Pt/CNT-based electrode revealed a
greatly reduced mass transport overpotential compared with the Pt/VXC72R-based
electrodes as shown in their polarization curves.


Fig. 4.10 Polarization curves of Pt/CNT and Pt/VXC72R-based
electrodes with 0.04 mg cm
-2
Pt at cathode.

105














Figure 4.11 shows the in situ CVs of Pt/CNT and Pt/VXC72R-based electrodes
at a scan rate of 100 mV s
-1
. The typical characteristics of the cyclic voltammogram
for Pt metal observed for both catalysts are a pair of hydrogen adsorption/desorption
peaks located below 0.3 V and a pair of Pt oxidation/reduction peaks at around 0.8 V
[28]. It is noteworthy that the Pt oxidation peak of the Pt/CNT catalyst showed a
positive shift to a higher potential compared with the Pt/VXC72R catalyst, which
implies that the sputter-deposited Pt catalyst on the CNT support may have better
electrochemical stability over the commercial catalyst. In addition, the cyclic
voltammogram for the Pt/CNT catalyst revealed a notably smaller double-layer
charging region than the commercial catalysts, which is attributed to the high
electronic conductivity of the CNT support as proposed by Xing [29]. The in situ
electrochemical active surface area (ECSA) was obtained for the three electrodes by
0.00.20.40.60.81.01.2

-200
-150
-100
-50
0
50
100
150
200


Current density / mA mgPt
-1
Potential / V
Pt/CNT catalyst
20 wt% Pt/VXC72R catalyst (E-TEK)
40 wt% Pt/VXC72R catalyst (Johnson Matthey)
Fig. 4.11 In situ CVs of Pt/CNT and Pt/VXC72R-based
electrodes with 0.04 mg cm
-2
Pt at cathode.
106

calculating the hydrogen adsorption/desorption peak areas excluding the double-layer
capacitance region. The average of the cathodic and anodic peak areas was used as the
total charge transfer due to hydrogen adsorption/desorption (Q
H
) in our study. The
ECSA can be calculated from S
EC

= Q
H
/(Q
ref
× (Pt loading)), where Q
ref
= 210 µC
cm
-2
, corresponding to the charge per area of Pt with monolayer adsorption of
hydrogen [11]. The calculated in situ ECSA for the Pt/CNT-based electrode was 17.2
m
2
g
-1
, in comparison to 29.1 m
2
g
-1
for the 20wt% Pt/VXC72R-based electrode and
15.6 m
2
g
-1
for the 40wt% Pt/VXC72R-based electrode. The highest ECSA observed
from the 20wt% Pt/VXC72R-based electrode may account for its well-dispersed
Nafion ionomers throughout the catalyst layer. The Pt/CNT-based electrode also
showed relatively high ECSA owing to its distinct morphological property that all the
Pt nanoparticles are highly localized at the electrode-electrolyte interface by direct
sputter-deposition. On the other hand, the Pt/VXC72R catalysts were reported to have

a rather dense structure, where the three-phase reaction zone was greatly impaired by
the compact carbon particles. Therefore, it is likely that the Pt/CNT-based electrode
should give higher Pt utilization than the Pt/VXC72R-based electrodes, whereas
further investigation on these three electrodes in addition to the ECSA evaluation is
necessary to provide more insights into the notable cell performance improvement of
the Pt/CNT-based electrode.

Subsequently, a more in-depth investigation on the Pt/CNT-based electrode was
conducted using a.c. impedance spectroscopy, to further understand the underlying
mechanisms for the performance improvement from this Pt/CNT layer.
Electrochemical impedance spectroscopy (EIS) was used in that it is a reliable
diagnostic tool for evaluating the transport properties in fuel cells owing to its ability
107

to separate the impedance responses of the various transport processes occurring
simultaneously in PEMFCs [30]. A number of EIS simulation studies on working
PEMFCs have suggested several in situ characteristics of a PEMFC MEA present in
its impedance spectra [31-34]: (i) the cathode impedance predominates in the
spectrum while the anode impedance is relatively negligible at low overpotential; (ii)
the charge transfer resistance (R
ct
) of the cathode due to the oxygen reduction reaction
(ORR), in parallel with the double-layer capacitance (C
dl
) owing to the porous
structure of the electrode; (iii) the ohmic resistance (R
ohm
) distributed in the electrode
mainly caused by the electrolyte membrane; (iv) the oxygen diffusion resistance (R
od

)
derived from the diffusion limitation of oxygen in the cathode, especially when air is
used as oxidant; (v) the water transport resistance (R
wt
) as a result of water transport
limitation in the membrane. Generally, the high frequency region of the impedance
spectrum reflects the charge transfer properties of the catalyst layer, whereas the low
frequency region represents the mass transport properties in the electrode.
Consequently, EIS enables us to evaluate them separately in order to determine their
individual influence on the overall fuel cell performance.

In order to distinguish the contributions of different processes that take place in
the Pt/CNT and Pt/VXC72R-based electrodes, in situ EIS tests were performed to
measure the frequency dependence of the impedance of each electrode at different
overpotential regions. The EIS results of the three electrodes are presented in Nyquist
plot as shown in Fig. 4.12. At 0.8 V, which is near open circuit voltage, the rate
determining step is controlled by the kinetics of ORR and thus the major impedance
of all the MEAs arises from the charge transfer resistance R
ct
and double-layer
capacitance C
dl
in the cathode catalyst layer observed as a semicircular arc in Fig.