Journal of Colloid and Interface Science 362 (2011) 607–614
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science
www.elsevier.com/locate/jcis
Hydrophilicity/porous structure-tuned, SiO2/polyetherimide-coated polyimide
nonwoven porous substrates for reinforced composite proton exchange membranes
Jung-Ran Lee a, Ji-Hye Won a, Na-Young Kim b, Moo-Seok Lee b, Sang-Young Lee a,⇑
a
b
Department of Chemical Engineering, Kangwon National University, Chuncheon, Kangwondo 200-701, Republic of Korea
Eco Research Institute, Kolon Central Research Park, Yongin, Kyunggido 446-797, Republic of Korea
a r t i c l e
i n f o
Article history:
Received 17 March 2011
Accepted 26 June 2011
Available online 7 July 2011
Keywords:
Polymer electrolyte membrane fuel cells
Reinforced composite membranes
Porous substrates
Silica
Polyimide nonwovens
Nafion
a b s t r a c t
Porous substrate-reinforced composite proton exchange membranes have drawn considerable attention
due to their promising application to polymer electrolyte membrane fuel cells (PEMFCs). In the present
study, we develop silica (SiO2) nanoparticles/polyetherimide (PEI) binders-coated polyimide (PI) nonwoven porous substrates (referred to as ‘‘S-PI substrates’’) for reinforced composite membranes. The properties of S-PI substrates, which crucially affect the performance of resulting reinforced composite
membranes, are significantly improved by controlling the hygroscopic SiO2 particle size. The 40 nm SPI substrate (herein, 40 nm SiO2 particles are employed) shows the stronger hydrophilicity and highly
porous structure than the 530 nm S-PI substrate due to the larger specific surface area of 40 nm SiO2 particles. Based on the comprehensive understanding of the S-PI substrates, the structures and performances
of the S-PI substrates-reinforced composite membranes are elucidated. In comparison with the 530 nm SPI substrate, the hydrophilicity/porous structure-tuned 40 nm S-PI substrate enables the impregnation of
a large amount of a perfluorosulfonic acid ionomer (Nafion), which thus contributes to the improved proton conductivity of the reinforced Nafion composite membrane. Meanwhile, the reinforced Nafion composite membranes effectively mitigate the steep decline of proton conductivity with time at low
humidity conditions, as compared to the pristine Nafion membrane. This intriguing finding is further discussed by considering the unusual features of the S-PI substrates and the state of water in the reinforced
Nafion composite membranes.
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
Proton exchange membranes are considered a critical component that can facilitate the successful commercialization of polymer electrolyte membrane fuel cells (PEMFCs). In particular,
when electric vehicle applications necessitating high-power density and cost competitiveness are targeted, thin and high-conductive polymer electrolytes are strongly demanded. A formidable
challenge in developing the proton exchange membranes is the
accompanying loss of mechanical strength and dimensional
change [1–9]. Among various approaches to overcome these limitations, reinforced composite membranes consisting of mechanically reinforcing porous substrates and polymer electrolytes have
attracted considerable attention. Representative examples of the
porous substrates include poly(tetrafluoroethylene) (PTFE) [1–3],
polycarbonate [4], polypropylene [5], glass paper [6], and polyimide [7,8]. Porous structure, compatibility with filling electrolytes,
and chemical/mechanical stability of porous substrates are crucial
⇑ Corresponding author. Fax: +82 33 251 3658.
E-mail address: syleek@kangwon.ac.kr (S.-Y. Lee).
0021-9797/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcis.2011.06.076
factors affecting the performance of reinforced composite membranes [3–8].
As part of ongoing research to fulfill the aforementioned
requirements, we have developed a new reinforcing porous substrate by exploiting a unique concept of silica (SiO2) particlescoated polyimide (PI) nonwovens. We previously reported that
the introduction of hygroscopic SiO2 particles interconnected by
polyetherimide (PEI) binders effectively improves not only the
mechanical strength but also the polarity of the PI nonwoven [9].
A perfluorosulfonic acid ionomer (Nafion) was successfully
impregnated into the SiO2/PEI-coated PI nonwoven substrate and
the resulting reinforced composite membrane remarkably suppressed the dimensional change and retarded the decline of proton
conductivity at low humidity conditions.
In the present study, as a continued effort, we focus on further
enhancing the properties of SiO2/PEI-coated PI nonwoven substrates
(hereinafter, referred to as ‘‘S-PI substrates’’) by tuning SiO2 powder
size. SiO2 particles of different sizes (40 and 530 nm) are employed
and their influences on the hydrophilicity and porous structure of
S-PI substrates are investigated. Based on comprehensive understanding of the S-PI substrates, the structures and performances of
the S-PI substrates-reinforced Nafion composite membranes
608
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
(hereinafter, referred to as ‘‘composite membranes’’) are elucidated
as a function of SiO2 particle size. Meanwhile, it is known that when
water-swollen membranes such as Nafion are dehydrated, they may
face serious difficulties in preserving proton conductivity [10–13],
because their proton transport depends heavily on degree of hydration. Time evolution of the proton conductivity of composite membranes is examined under a harsh operating condition of 20%
relative humidity (RH), and the results are discussed in great detail
with the consideration of the unusual features of the S-PI substrates
and the state of water in the composite membranes.
2. Experimental
2.1. Preparation of SiO2/PEI-coated PI nonwoven substrates and
reinforced Nafion composite membranes
Electrospun PI nonwovens (KFM-NTÒ, average thickness =
17 lm, average porosity = 77%) were provided from Kolon (Korea).
Two different SiO2 powders (average particle size (average specific
surface area) = 40 nm (30 m2 g1) and 530 nm (6 m2 g1)) were obtained from Denki Kagaku (Japan). SiO2 coating solutions were prepared by mixing SiO2 powders and amorphous thermoplastic PEI
(ULTEM 1000) in dimethylacetamide (DMAc) as a solvent, followed
by additional mixing via bead-milling for 2 h. The SiO2/PEI coating
solutions were then applied to the PI nonwovens via a dip-coating
process. The coating solution-immersed PI nonwovens were subsequently dried at 80 °C for 2 h to evaporate DMAc and then further
vacuum dried at 80 °C for 2 h. The thickness of the SiO2/PEI-coated
PI nonwoven substrates (i.e., S-PI substrates) was observed to be
around 18 lm (Table 1).
The SiO2/PEI ratio was fixed at 95/5 (wt%/wt%), wherein influence of the PEI binder on the S-PI substrates is considered to be
insignificant due to its very small amount, as compared to the
SiO2 content [9]. The SiO2/PEI contents in the coating solutions
were 5 wt% for 40 nm SiO2 and 20 wt% for 530 nm SiO2, respectively. Meanwhile, in order to conduct a direct comparison between the 40 nm SiO2 and 530 nm SiO2, 5 wt% of 530 nm SiO2/
PEI coating solution was also prepared and applied to the PI nonwoven. Unfortunately, the resulting 530 nm S-PI substrate (from
the 5 wt% of 530 nm SiO2/PEI coating solution) showed the excessively poor hydrophilicity. This made it difficult for the 530 nm S-PI
substrate to be impregnated with Nafion, which consequently limited further characterization of reinforced Nafion composite membrane. Hence, our interest was focused on comparison of the 5 wt%
of 40 nm SiO2/PEI coating solution with the 20 wt% of 530 nm SiO2/
PEI coating solution.
Meanwhile, for fabrication of the reinforced Nafion composite
membranes (i.e., composite membranes), the S-PI substrates were
extended on a glass frame and immersed in a Nafion DE520 solution (solid content = 5 wt%, DuPont). The Nafion solution-soaked
S-PI substrates were then dried at 130 °C for 5 min. The impregnation and drying steps were repeated five times in order to eliminate voids and pinholes in the composite membranes. The
composite membranes were then acidified in a 1 N boiling sulfuric
Table 1
Basic characteristics of PI nonwoven, 530 nm S-PI substrate, and 40 nm S-PI substrate.
Thickness (lm)
Porosity (%)
Actual surface area of SiO2
particles (I) (m2)
Actual surface area of SiO2
particles (II) (m2)
acid solution for 2 h and subsequently rinsed with distilled water
for 2 h. The thickness of the composite membranes was approximately 40 lm.
2.2. Characterization of SiO2/PEI-coated PI nonwoven substrates and
reinforced Nafion composite membranes
The surface and cross-sectional morphologies of the S-PI substrates and the composite membranes were investigated using a
field emission scanning electron microscope (FE-SEM, S-4300, Hitachi) equipped with an energy-dispersive spectrometer (EDS). The
pore size and pore size distribution of the S-PI substrates were
quantitatively determined by a bubble-point test performed to
the ASTM standard F316 using a capillary flow porometer
(CFP-1500AE, PMI). The porosity of the S-PI substrates, Up (%),
was estimated using the following equation [14,15]:
Up ð%Þ ¼ f1 ½ðW C =qC þ W N =qN Þ=V SN g 100
ð1Þ
where WC is the weight per square meter of SiO2/PEI coating layer
(=3.18 g for the 40 nm S-PI substrate and 7.56 g for the 530 nm
S-PI substrate), WN is the weight per square meter of the PI nonwoven (=4.47 g), qC is the density of the SiO2/PEI coating layer
(=2.12 g cc1), qN is the density of the PI nonwoven (=1.37 g cc1),
and VSN is the volume of the S-PI substrate (=1.8 105 m3).
For measurement of the dimensional change (i.e. area-based
dimensional expansion), the membranes were soaked in deionized
water overnight at room temperature. The specimens were then
dried in a vacuum oven at 80 °C for 24 h. The area of the specimens
was recorded before (Awet) and after (Adry) the vacuum drying step.
The change in area (DF) was calculated using the following
equation:
DA ð%Þ ¼ ½ðAwet Adry Þ=Adry 100
ð2Þ
The temperature-dependent proton conductivities of the membranes at 100% RH were estimated with an impedance analyzer
(VSP classic, Bio-Logic) using a four-probe method over a frequency range of 101–106 Hz [9,16,17]. Meanwhile, in order to
investigate the proton conductivity of the membranes at low RH
conditions, the membranes were placed in a temperature/humidity
control chamber (SH-241, ESPEC) under a given temperature (30
and 65 °C) and 20% RH. The variation in their proton conductivity
was then measured as a function of elapsed time. The state of
water in the membranes was examined by a thermogravimetric
analyzer (TGA, SDT Q600, TA Instruments) [9,18,19]. The TGA
experiment was performed in a temperature range from room
temperature (via isothermal heating at 100 °C for 40 min) to
400 °C at a heating rate of 10 °C min1 under a nitrogen atmosphere. Prior to the TGA measurement, the membranes were
immersed in water at 30 °C for 8 h and then preequilibrated at
30 °C and 50% RH for 2 h. The fraction of physically adsorbed water
was quantified by measuring the weight loss below 100 °C. The
amount of chemically adsorbed water was calculated by subtracting the physically adsorbed water content from the total water
content. In order to obtain the total water content in the membranes, the samples preequilibrated at 30 °C and 50% RH for 2 h
were vacuum-dried at 100 °C for 12 h. The total water content
(DW) was calculated by the weight difference of the samples
before and after the vacuum drying step:
PI
nonwoven
530 nm S-PI
substrate
40 nm S-PI
substrate
DW ðwt%Þ ¼ ½ðW wet W dry Þ=W wet 100
17
77
–
18
62
41
18
73
103
3. Results and discussion
–
43
91
ð3Þ
The basic characteristics of the S-PI substrates were first
investigated. Fig. 1a shows that the pristine PI nonwoven has
609
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
40
Population of Pores (%)
530nm SiO2 /PEI-coated PI nonwoven
40nm SiO2 /PEI-coated PI nonwoven
PI nonwoven
30
20
10
0
0.5
1.0
1.5
2.0
Pore Diameter (um)
2.5
3.0
Fig. 2. Pore size distribution of pristine PI nonwoven, 530 nm S-PI substrate (SiO2/
PEI content in coating solution = 20 wt%), and 40 nm S-PI substrate (SiO2/PEI
content in coating solution = 5 wt%).
Fig. 1. FE-SEM photographs (surface morphology) of: (a) pristine PI nonwoven, (b)
530 nm S-PI substrate (SiO2/PEI content in coating solution = 20 wt%), and (c) 40 nm
S-PI substrate (SiO2/PEI content in coating solution = 5 wt%).
multi-fibrous layers with large pores formed between the PI fibers.
This is consistent with the results of previous publications [20,21]
reporting that electrospun nonwovens generally provide thin
thickness and also highly porous structure. Meanwhile, for the SPI substrates, it is observed that the SiO2 nanoparticles (both 530
and 40 nm) are successfully incorporated between the PI fibers
(Fig. 1b and c). An interesting finding is that due to the high content (=20 wt%) of 530 nm SiO2/PEI in the coating solution, a large
amount of SiO2/PEI (=7.56 g m2) is incorporated, which leads to
pore plugging in the 530 nm S-PI substrate. In contrast, for the
40 nm S-PI substrate, a relatively small amount of SiO2/PEI
(=3.18 g m2) is coated onto the PI fibers and appears to have insignificant influence on the overall porous structure of the pristine PI
nonwoven.
This intriguing effect of SiO2 particle size on the porous structure of the S-PI substrates was quantitatively identified in terms
of the pore size and pore size distribution. Fig. 2 shows that,
whereas most pores of the pristine PI nonwoven are in a range of
1.5–3.0 lm, the 530 nm S-PI substrate presents a high portion of
small-sized pores (in the range of 0.1–1.0 lm) due to the incorporation of the large amount of 530 nm SiO2/PEI. On the other hand,
the pore size and pore size distribution (approximately, 1.3–
2.7 lm) of the 40 nm S-PI substrate appear to be similar to those
of the pristine PI nonwoven.
The aforementioned porous structure of S-PI substrates was further elucidated by measuring their porosities. Table 1 shows that
in comparison with the pristine PI nonwoven (porosity = 77%),
the S-PI substrates present slightly lower porosities, i.e. 62% for
530 nm S-PI substrate and 73% for 40 nm S-PI substrate. This reveals that the 40 nm S-PI substrate still maintains a highly porous
structure after being coated with the SiO2/PEI. As a consequence,
the 40 nm S-PI substrate is advantageous in impregnating a large
amount of Nafion, in comparison with the 530 nm S-PI substrate.
The interfacial affinity between porous substrates and proton
conducting electrolytes is known to critically affect the preparation
of porous substrate-reinforced composite membranes [1–9]. For
instance, for a polytetrafluoroethylene (PTFE) porous substrate,
one of the most widely used reinforcing substrates, the fabrication
of PTFE-reinforced Nafion composite membranes without voids or
pinholes is very challenging, due to the difficulties associated with
the impregnation of hydrophilic Nafion solution into pores of the
hydrophobic PTFE substrate. Most water-swollen proton conducting polymers tend to be hydrophilic and are dispersed/dissolved
in polar solvents. Hence, a reinforcing porous substrate with high
polarity should be preferred with respect to the impregnation of
polar proton conducting electrolytes. Fig. 3 depicts the wetting
behavior of the various porous substrates toward Nafion solution
(concentration = 10 wt%) and also water. It is clearly observed that,
because of the hydrophobic nature of the PI, the pristine PI nonwoven does not readily absorb the hydrophilic Nafion solution
(Fig. 3a). In contrast, the S-PI substrates show the better wettability
with the Nafion solution due to the presence of hygroscopic SiO2
particles. This demonstrates that the incorporation of hygroscopic
SiO2 particles is an effective approach for modifying the PI nonwoven surface to be compatible with the Nafion solution, which is
therefore expected to enable facile impregnation of Nafion into
the S-PI substrates.
Meanwhile, as compared to the 530 nm S-PI substrate, the
40 nm S-PI substrate presents the superior affinity for the Nafion
solution even at its smaller SiO2 loading (i.e. 3.02 g m2 for
40 nm S-PI substrate vs. 7.18 g m2 for S-PI substrate). In order
to further elucidate the effect of SiO2 particle size on the Nafion
solution wettability, another 530 nm S-PI substrate was prepared
610
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
Fig. 3. Wettability of pristine PI nonwoven substrate, 530 nm S-PI substrate (SiO2/PEI content in coating solution = 5, 20 wt%), and 40 nm S-PI substrate (SiO2/PEI content in
coating solution = 5 wt%): (a) photographs after 10 wt% Nafion solution (or water) droplets falling and (b) quantitative comparison of water immersion height.
from a coating solution with SiO2/PEI content = 5 wt%, which is the
same SiO2/PEI content used for the fabrication of 40 nm S-PI substrate. Fig. 3a apparently shows that the 530 nm S-PI substrate
(SiO2/PEI content = 5 wt%) offers the very poor wettability toward
the Nafion solution, in comparison with the 40 nm S-PI substrate
(SiO2/PEI content = 5 wt%) as well as the 530 nm S-PI substrate
(SiO2/PEI content = 20 wt%).
In addition to the Nafion solution wettability, the water wettability of the nonwoven substrates was also examined (Fig. 3a). Both
the 40 nm S-PI substrate (SiO2/PEI content = 5 wt%) and 530 nm SPI substrate (SiO2/PEI content = 20 wt%) are easily soaked with
water, in contrast to the pristine PI nonwoven and 530 nm S-PI
substrate (SiO2/PEI content = 5 wt%). Notably, the 40 nm S-PI substrate shows the larger area coverage of water droplet than the
530 nm S-PI substrate (SiO2/PEI content = 20 wt%), which is consistent with the results of the Nafion solution wettability.
This improved water wettability of the 40 nm S-PI substrate is
further confirmed by carrying out a quantitative comparison of
water immersion height between the nonwoven substrates
(Fig. 3b), wherein a certain amount (1.4 wt%) of red ink was
added into water in order to provide clear images. After an elapsed
time of 10 min, the 40 nm S-PI substrate (SiO2/PEI content = 5 wt%)
presents the higher water immersion height (1.2 cm vs. 0.7 cm)
than the 530 nm S-PI substrate (SiO2/PEI content = 20 wt%). On the
other hand, the water immersion height of the 530 nm S-PI substrate (SiO2/PEI content = 5 wt%) as well as the PI nonwoven is
too low to be accurately determined.
The effect of SiO2 particle size on the wettability of S-PI substrates was further examined by investigating the actual surface
area of SiO2 particles per unit area (=1.0 m2) of S-PI substrates,
wherein the 530 nm S-PI substrate (SiO2/PEI content = 20 wt%)
and 40 nm S-PI substrate (SiO2/PEI content = 5 wt%) are considered. Under the assumption that SiO2 particles are spherical and
have a uniform particle size distribution, the actual surface area
of SiO2 particles per unit area of S-PI substrates can be calculated
from the SiO2 content (=7.18 g m2 for 530 nm S-PI substrate vs.
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
3.02 g m2 for 40 nm S-PI substrate) and SiO2 density
(=2.20 g cc1). This calculation presents that the actual surface
areas of SiO2 particles per unit area of S-PI substrates (actual surface area of SiO2 particles (I) shown in Table 1) are, respectively,
41 m2 for the 530 nm S-PI substrate and 103 m2 for the 40 nm SPI substrate. Meanwhile, in the experimental section, it was already informed from the SiO2 supplier that the 40 nm SiO2 particles have an average specific surface area of 30 m2 g1, whereas
the 530 nm SiO2 particles show an average specific surface area
of 6 m2 g1. By exploiting this basic information of SiO2 powders,
actual surface area of SiO2 particles per unit area of S-PI substrates
(actual surface area of SiO2 particles (II) shown in Table 1) can be
also obtained, which yields 6 m2 g1 7.18 g = 43 m2 for the
530 nm S-PI substrate and 30 m2 g1 3.02 g = 91 m2 for the
40 nm S-PI substrate. It is apparent that both the aforementioned
results of actual surface area of SiO2 particles per unit area of SPI substrates are close to each other and also confirm that the
40 nm SiO2 particles provide the larger actual surface area per unit
area of S-PI substrate than the 530 nm SiO2 particles even at the
small SiO2 loading. This demonstrates that the incorporation of
the small-sized (40 nm) SiO2 particles is more effective in rendering the surface of the PI nonwoven hydrophilic and compatible
with the Nafion solution, in comparison with the large-sized
(530 nm) SiO2 particles.
Based on the aforementioned characterization of the S-PI substrates, the structure and performance of the composite membranes were investigated. The cross-sectional morphology (Fig. 4)
verifies the presence of the S-PI substrates in the composite membranes. Meanwhile, the pristine PI nonwoven-reinforced Nafion
composite membrane was also prepared; however, when it was
subjected to acidification, most of the Nafion was detached from
the pristine PI nonwoven substrate. More detailed explanation on
this adhesion failure was presented in a previous publication [9].
Hence, characterization of the pristine PI nonwoven-reinforced
Nafion composite membrane could not be carried out. This indicates that the interfacial adhesion between the reinforcing porous
substrates and Nafion is critical to the fabrication of reinforced
composite membranes, and thus, the introduction of hygroscopic
611
SiO2 particles to PI nonwovens can be a promising approach to enhance the interfacial compatibility.
Fig. 5 compares the dimensional change (i.e. area-based dimensional expansion) of the composite membranes with that of the
Nafion membrane. The Nafion membrane (thickness = ca. 40 lm)
was prepared by employing the same Nafion solution and fabrication procedure (i.e. casting the Nafion solution onto a glass plate,
followed by drying at 130 °C) used in the preparation of the composite membrane. The two composite membranes present the
noticeably suppressed dimensional change (DA < 5%), whereas
the Nafion membrane shows the large dimensional change
(DA 45%). This improvement in the dimensional change reveals
that the inclusion of the S-PI substrates in the composite membranes is effective in suppressing the swelling of the impregnated
Nafion. This finding is consistent with the results of previous studies on porous substrate-reinforced composite membranes [1–9].
The proton conductivity of the composite membranes is investigated as a function of SiO2 particle size. Fig. 6 shows that at a
condition of 100% RH, the temperature-dependent proton conductivities of the membranes increase as the measurement temperature is increased from 30 to 80 °C. Meanwhile, over a wide range
of temperatures, the composite membranes show lower proton
conductivity than the Nafion membrane. This decrease in the proton conductivity of the composite membranes is ascribed to the
presence of the S-PI substrates that are inert to proton conduction.
Interestingly, the 40 nm S-PI substrate-reinforced composite membrane offers higher proton conductivity (0.07 (at 30 °C)–0.12 (at
80 °C) S cm1) than the 530 nm S-PI substrate-reinforced composite membrane (0.05 (at 30 °C)–0.09 (at 80 °C) S cm1). This difference in the proton conductivity between the composite
membranes could be explained by considering the porous structure and porosity of the S-PI substrates. It was already shown in
Figs. 1 and 2 that the 40 nm S-PI substrate provides the highly porous structure than the 530 nm S-PI substrate. In general, proton
conductivity of porous substrate-reinforced composite membranes
is strongly dependent on the porous structure of reinforcing porous
substrates [4–9], because proton transport occurs exclusively in
pore-impregnating electrolytes. In this regard, the aforementioned
Fig. 4. FE-SEM photographs (cross-sectional morphology) of reinforced Nafion composite membranes incorporating: (a) 530 nm S-PI substrate and (b) 40 nm S-PI substrate.
612
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
Fig. 5. Dimensional change of: (a) Nafion membrane, (b) reinforced Nafion
composite membrane incorporating 530 nm S-PI substrate, and (c) reinforced
Nafion composite membrane incorporating 40 nm S-PI substrate, where (areabased) dimensional change (DF) is calculated by the equation of DF
(%) = [(Awet Adry)/Adry] 100.
-1
Proton Conductivity (S cm )
0.14
0.12
Fig. 7. Schematic illustrations of structure and conceptual proton transport
pathway of reinforced Nafion composite membranes incorporating: (a) 530 nm SPI substrate and (b) 40 nm S-PI substrate.
0.10
0.08
0.06
Nafion Membrane
0.04
Reinforced Nafion Composite
Membtane (40nm SiO2 )
0.02
0.00
Reinforced Nafion Composite
Membtane (530nm SiO2)
30
40
50
60
70
80
o
Temperature ( C)
Fig. 6. Temperature-dependent proton conductivity of Nafion membrane and
reinforced Nafion composite membranes at a condition of 100% RH.
comparison of proton conductivity between the 40 nm S-PI substrate and 530 nm S-PI substrate demonstrates that the proton
conductivity of the composite membranes can be improved by tuning the porous structure of reinforcing substrates. Schematic illustrations of the structure and conceptual proton transport pathway
of the composite membranes incorporating 530 nm S-PI substrate
or 40 nm S-PI substrate are given in Fig. 7.
In addition to the proton conductivity at 100% RH, the proton
conductivity of membranes at a low humidity condition of 20%
RH was examined under given temperatures (30 and 65 °C).
Fig. 8a shows that at a condition of 30 °C/20% RH, the proton conductivity of the Nafion membrane dramatically decreases as the
measurement time is increased. The proton conductivity of
water-swollen membranes such as Nafion is known to depend
strongly on water content in the membranes [11–13]. Therefore,
this decrease in the proton conductivity reveals that the membranes are being dehydrated under the low humidity condition.
On the other hand, the composite membranes show the different
proton conductivity behavior. In the early stage of the conductivity
measurement, due to the inactive volume occupied by the S-PI
substrates, the composite membranes present lower proton conductivity than the Nafion membrane. However, in contrast to the
Nafion membrane, the decrease in proton conductivity with time
is heavily retarded. More notably, after elapsed time of about
30 min (for the 40 nm S-PI substrate) and 70 min (for the 530 nm
S-PI substrate), the composite membranes deliver higher proton
conductivity than the Nafion membrane.
This interesting result is further confirmed by measurement of
proton conductivity at a harsher condition of 65 °C/20% RH
(Fig. 8b). In comparison with the condition of 30 °C/20% RH, the
drop in proton conductivity is much steeper, and thus after shorter
elapsed time, the membranes completely lose their proton conducting capability. However, even at this severe condition, the
composite membranes are still effective in mitigating the loss of
proton conductivity. This improvement in the proton conductivity
at the low humidity conditions is attributed to the presence of
hygroscopic SiO2 nanoparticles in the reinforcing porous substrates, which can tightly retain water molecules in the composite
membranes. It is well known that hygroscopic powders such as
SiO2, TiO2, and ZrO2 have high affinity for water molecules, which
consequently contributes to strong retention of water molecules
[9,22–24].
In order to obtain a better understanding of the proton conductivity at the dehumidified conditions, the state of water in the
membranes was investigated in detail. Previous studies [9,18,19]
reported that the state of water in water-swollen membranes
could be classified into two groups based on an analysis of TGA
thermograms: physically adsorbed water (i.e., free and slightly
bound water) and chemically adsorbed water (i.e., strongly bound
613
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
0.08
Reinforced Nafion Composite
Membrane (530nm SiO2 )
200
90
100
Nafion Membrane
0.06
80
10
30
40
50
Time (min)
100
0.04
0.02
0.00
0
60
120
180
240
Time (min)
Physically Chemically
Adsorbed Adsorbed
90
Naf
Nafion Membrane
Reinfoced Nafion Composite
Membrane (530nm SiO 2 )
0.10
Water
Reinforced Nafion Composite
Rei
Mem
Membrane (40nm SiO2 )
Reinforced Nafion Composite
Membrane (40nm SiO2)
0.12
Water
Rei
Reinforced
Nafion Composite
Membrane (530nm SiO2 )
Mem
(b) 0.14
80
100
Nafion Membrane
200
400
Temperature ( C)
0.08
(a)
0.06
10
Physically Adsorbed Water
Chemicaly Adsorbed Water
0.04
0.02
0.00
300
o
0
3
6
9
12
15
18
21
Time (min)
Fig. 8. Variation in proton conductivity of Nafion membrane and reinforced Nafion
composite membranes as a function of elapsed time: (a) at 30 °C/20% RH and (b) at
65 °C/20% RH.
water). More specifically, the weight loss below 100 °C can be
attributed to the vaporization of physically adsorbed water in
membranes, and the weight loss above 100 °C (=total water content in membranes the weight loss below 100 °C) can be ascribed
to the desorption of chemically adsorbed water. Meanwhile,
Zawodzinski et al. [25] raised a cautionary note on the quantitative
use of thermodynamic data for elucidating the state of water in
water-swollen membranes. Thus, in the present study, the characterization of water molecules based on the TGA thermograms has
been restricted to carrying out a relative comparison between
the samples, rather than providing the absolute value of the state
of water.
Fig. 9a depicts the weight loss profile of water for the Nafion
membrane and composite membranes. The inset in Fig. 9a indicates the weight loss profile of water at an isothermal condition
of 100 °C/40 min, which was introduced to allow sufficient time
for evaporation of physically adsorbed water [9]. By exploiting
these TGA results, the elucidation of physically adsorbed water
and chemically adsorbed water in the total water content was carried out. Fig. 9b shows that the total water content is 5.66 wt% for
the Nafion membrane, 5.11 wt% for the 530 nm S-PI substratereinforced composite membrane, and 6.28 wt% for the 40 nm S-PI
substrate-reinforced composite membrane, respectively. More
specifically, in the Nafion membrane, the weight loss for physically
adsorbed water is 4.60 wt% and the weight loss for chemically adsorbed water is 1.06 wt%. On the other hand, in the composite
membranes, the results of weight loss for physically adsorbed
water vs. the weight loss for chemically adsorbed water are
Water Content (wt%)
Proton Conductivity (S cm-1)
20
Temperature (oC)
Weight (%)
Reinforced Nafion Composite
Membrane(40nm SiO2)
Weight (%)
Proton Conductivity (S cm-1)
300
100
(a) 0.10
8
6
3.35
4
2.35
4.60
2
0
2.76
2.93
Composite
(530 nm SiO2)
Composite
(40 nm SiO2)
1.06
Nafion
Membrane
(b)
Fig. 9. Characterization of state of water for Nafion membrane and reinforced
Nafion composite membranes: (a) TGA thermograms at a heating rate of
10 °C min1, wherein the inset indicates the weight loss profile of water at an
isothermal condition of 100 °C/40 min and (b) analysis of physically adsorbed water
and chemically adsorbed water in the total water content of the membranes.
2.35 wt% vs. 2.76 wt% (for the 530 nm S-PI substrate) and
3.35 wt% vs. 2.93 wt% (for the 40 nm S-PI substrate). This TGA characterization confirms that the composite membranes have a larger
amount of chemically adsorbed water due to the presence of
hygroscopic SiO2 particles in the reinforcing porous substrates.
In dehumidified conditions, proton conduction is predominantly governed by the vehicle mechanism rather than the
Grotthus mechanism [26–28]. Under these circumstances, the contribution of chemically adsorbed water to proton transport becomes more significant than that of physically adsorbed water.
Therefore, the increase in the chemically adsorbed water content
in the composite membranes could be a strong evidence to account
for the retarded loss of proton conductivity at a low humidity condition of 20% RH. These results on the proton conductivity and the
state of water emphasize that the S-PI substrates can be exploited
to facilitate proton transport of the composite membranes at dehumidified conditions.
614
J.-R. Lee et al. / Journal of Colloid and Interface Science 362 (2011) 607–614
4. Conclusions
References
We have significantly improved the hydrophilicity and porous
structure of S-PI substrates by controlling the SiO2 particle size
(herein, 40 and 530 nm). The 40 nm S-PI substrate, even at its
smaller SiO2 loading, offered the stronger hydrophilicity than the
530 nm S-PI substrate due to the larger specific surface area of
40 nm SiO2 particles. This improved hydrophilicity of the 40 nm
S-PI substrate enabled the development of highly porous structure,
which thus facilitated the impregnation of a larger amount of Nafion. As a result, the hydrophilicity/porous structure-tuned 40 nm
S-PI substrate endowed the reinforced Nafion composite
membrane with the higher temperature-dependent proton conductivity, in comparison with the 530 nm S-PI substrate. The
hygroscopic SiO2 particles of the S-PI substrates contributed to
the increase in the chemically adsorbed water content in the composite membranes. Hence, the composite membranes effectively
suppressed the decline of proton conductivity with time at a harsh
condition of 20% RH, as compared to the Nafion membrane. The
present study underlines that the SiO2 particle size plays a crucial
role in tailoring the hydrophilicity and porous structure of the S-PI
substrates, which significantly influence the performance of resulting composite membranes. Our future studies will be devoted to
electrochemical characterization of MEA (membrane-electrode
assemblies) assembled with composite membranes.
[1] F. Liu, B. Yi, D. Xing, J. Yu, H. Zhang, J. Membr. Sci. 212 (2003) 213.
[2] L. Wang, B.L. Yi, H.M. Zhang, Y.H. Liu, D.M. Xing, Z.G. Shao, Y.H. Cai, J. Power
Sources 167 (2007) 47.
[3] H. Tang, M. Pan, F. Wang, P.K. Shen, S.P. Jiang, J. Phys. Chem. B. 111 (2007)
8684.
[4] M.P. Rodgers, J. Berring, S. Holdcroft, Z. Shi, J. Membr. Sci. 321 (2008) 100.
[5] K.H. Kim, S.Y. Ahn, I.H. Oh, H.Y. Ha, S.A. Hong, M.S. Kim, Y.K. Lee, Y.C. Lee,
Electrochim. Acta 50 (2004) 577.
[6] T. Tezuka, K. Tadanaga, A. Matsuda, A. Hayashi, M. Tatsumisago, Electrochem.
Comm. 7 (2005) 245.
[7] H. Munakata, D. Yamamoto, K. Kanamura, Chem. Comm. (2005) 3986.
[8] T. Nguyen, X. Wang, J. Power Sources 195 (2010) 1024.
[9] J.R. Lee, N.Y. Kim, M.S. Lee, S.Y. Lee, J. Membr. Sci. 367 (2011) 265.
[10] K. Tasaki, R. DeSousa, H. Wang, J. Gasa, A. Venkatesan, P. Pugazhendhi, R.O.
Loutfy, J. Membr. Sci. 281 (2006) 570.
[11] G. Alberti, M. Casciola, Solid state protonic conductors, Solid State Ionics 145
(2001) 3.
[12] K.A. Mauritz, R.B. Moore, Chem. Rev. 104 (2004) 4535.
[13] K.D. Kreuer, J. Membr. Sci. 185 (2001) 29.
[14] T. Yamaguchi, F. Miyata, S. Nakao, J. Membr. Sci. 214 (2003) 283.
[15] T. Yamaguchi, H. Kuroki, F. Miyata, Electrochem. Commun. 7 (2005) 730.
[16] K.S. Yoon, J.H. Choi, Y.T. Hong, S.K. Hong, S.Y. Lee, Electrochem. Comm. 11
(2009) 1492.
[17] S.Y. So, Y.T. Hong, S.C. Kim, S.Y. Lee, J. Membr. Sci. 346 (2010) 131.
[18] S.J. Kim, C.K. Lee, S.I. Kim, Characterization of the water State of hyaluronic
acid, poly(vinyl alcohol) interpenetrating polymer networks, J. Appl. Polym,
Sci. 92 (2004) 1467.
[19] S.P. Tung, B.J. Hwang, J. Membr. Sci. 241 (2004) 315.
[20] C. Yang, Z. Jia, Z. Guan, L. Wang, J. Power Sources 189 (2009) 716.
[21] H.R. Jung, D.H. Ju, W.J. Lee, X. Zhang, R. Kotek, Electrochim. Acta 54 (2009)
3630.
[22] Z. Jie, T. Haolin, P. Mu, J. Membr. Sci. 312 (2008) 41.
[23] S.Y. Chen, C.C. Han, C.H. Tsai, J. Huang, Y.W. Yang, J. Power Sources 171 (2007)
363.
[24] K.T. Park, U.H. Jung, D.W. Choi, K. Chun, H.M. Lee, S.H. Kim, J. Power Sources
177 (2008) 247.
[25] T.L. Kalapos, B. Decker, H.A. Every, H. Ghassemi, T.A. Zawodzinski, J. Power
Sources 172 (2007) 14.
[26] Y.S. Kim, L. Dong, M.A. Hickner, T.E. Glass, V. Webb, J.E. McGrath,
Macromolecules 36 (2003) 6281.
[27] M. Saito, N. Arimura, K. Hayamizu, T. Okdada, J. Phys. Chem. B 108 (2004)
16064.
[28] Y. Su, Y. Liu, Y. Sun, J. Lai, B. Liu, M.D. Guiver, J. Membr. Sci. 296 (2007) 21.
Acknowledgments
This work was supported by the New & Renewable Energy R&D
Program of the Korea Institute of Energy Technology Evaluation
and Planning (KETEP) Grant funded by the Korea government Ministry of Knowledge Economy (2008-N-FC12-J-01-2-100). This research was also supported by a Grant from the Fundamental R&D
Program for Core Technology of Materials funded by the Ministry
of Knowledge Economy. This work was also supported by the National Research Foundation of Korea Grant funded by the Korean
Government (MEST) (NRF-2009-C1AAA001-2009-0093307).