Conductivity and Electrochemical ORR Mass Transport Properties of Solid Polymer Electrolytes Containing Poly(styrene sulfonic acid) Graft Chains

Nafion 117 was manufactured by Du Pont de Nemours. Ethylenetetrafluoroethylene and membranes were provided by Cranfield University (UK). They were prepared by radiation grafting of styrene onto preformed ETFE (Du Pont) and polyvinylideneflouride (PVDF) (Nowofol) films and subsequently sulfonated, as previously described.²⁵ membranes were synthesized in this laboratory by a two-step method in which pseudo-living poly(sodium styrene sulfonate) macromonomer chains were emulsion copolymerized with styrene.²⁸ The procedure for the synthesis of membranes is described in the following sections.

Scheme I. Polymer structures of membranes employed in this work.

Sodium styrene sulfonate (SSNa) and 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) were used as received (Aldrich). Divinylbenzene (Aldrich) was washed with NaOH (aq.) 10% and distilled over under reduced pressure. Ethylene glycol was distilled under reduced pressure. Acrylonitrile (Aldrich) was purified by distillation under reduced pressure.

Scheme II. Synthesis of graft copolymers.

Three series of graft polymer membranes comprised of a polyacrylonitrile (PAN) backbone and graft chains of poly(sodium styrene sulfonate) (PSSNa) were prepared by a two-step method analogous to the synthesis of membranes previously reported.^{27 28}

The first step is a pseudo-living free radical polymerization of sodium styrene sulfonate (SSNa), which is terminated with divinylbenzene (DVB) to form macromonomers of PSSNa The synthetic procedure and characterization of these macromonomers have been described in detail elsewhere and summarized as follows.²⁸ An initiator system is injected into a solution of SSNa in an ethylene glycol/water mixture with a stable free radical agent (2,2,6,6-tetramethyl-1-piperidinyloxy). The solution is stirred and heated to reflux (120-125°C). The molecular weight of the polymer increases with polymerization time. When the reaction is complete, DVB is coupled to the polymer to provide a vinylic terminus for subsequent grafting. The degree of polymerization of the macromonomer is adjusted by varying the relative amounts of monomer to initiator used in the reaction. Three series of macromonomers with SSNa chain lengths of 16, 32, and 106 repeating units, and polydispersity indexes of were synthesized.

The second step is the synthesis of the graft polymer by means of an emulsion polymerization of acrylonitrile and The polymerization was carried out in deionized at 60°C. Polymers with different sulfonate content were obtained by varying the ratio of to acrylonitrile in the feed. These ratios were 3/3, 3/4.5, 3/6, 3/9, 3/12, and 3/18 acrylonitrile). The volume of aqueous solvent used was 30, 35, 40, 50, 70, and 90 mL, respectively. In a typical polymerization, 3.0 g of were dissolved in 27 mL of and the system was purged with 3.0 mL of acrylonitrile and 50 mg of in 3 mL of were added to the system. The solution was heated to 60°C with stirring for 3.5 h. The polymer was precipitated into ether/methanol (60/40 vol %) mixture, filtered, and air dried. To remove low molecular weight oligomers and homopolymers of the polymer was washed with cold water. After filtration the polymer was dried under vacuum. The copolymers are designated (#), where # represents the degree of polymerization (DP) of the PSSNa graft chain. Acidified polymers are designated (#), where PSSA represents polystyrene sulfonic acid.

Membranes were prepared by solution casting from dimethylsulfoxide (DMSO) solutions containing 2% Polymer suspensions having a concentration wt % were prepared at room temperature and heated to 150°C with stirring to form a clear solution. The solution was cast onto a glass plate and dried at 110°C for 10 h. A uniform, semitransparent film was released from the glass by immersion in water. Sodium ions were exchanged for protons by soaking the membranes in 0.5 M for 2 days, followed by rinsing and storing in deionized water.

To determine the ion exchange capacity (IEC) of membranes, samples in their acidic form were immersed in 50 mL of 2.0 M NaCl solution for 2 h and titrated with NaOH (0.025 M) to a phenolphthalein end point. After titration the membranes were placed for 1 h in 0.1 M HCl, rinsed with distilled water, and dried under vacuum at 80°C to constant weight. The membrane mass was measured using a Sartorius Basic BA21S balance. IEC was calculated using Eq. 1

The dimensions of the membranes were measured using a digital micrometer and caliper (Mitutoyo). Water contents were calculated as a volume percentage according to Eq. 2

Conductivity measurements were performed by ac impedance spectroscopy with a Hewlett-Packard 8753A network analyzer.²⁹ Measurements were performed on the acidic form of the membranes, stored in Milli-Q water for 24 h prior to use. Impedance spectra between 300 kHz and 1 GHz were obtained under ambient conditions using a coaxial probe (2 mm inner diameter (id) central disk surrounded by a ring of id 6.5 mm) to which a gold film was evaporated onto the surface. The conductivity obtained for Nafion 117 (0.06-0.08 S/cm) is consistent with published results.

Membranes were immersed in a saturated solution of for 24 h in order to enhance the contrast in electron density between ionic and nonionic domains. The samples were rinsed with and dried under vacuum. Samples were embedded in Spur resin (Canemco) and sectioned along the normal direction to yield thick slices using an ultramicrotome (Reichert OM U3). The slices were picked up with 100 mesh copper grids for TEM analysis. Micrographs were taken using a Zeiss microscope operated at 80 kV in the Electron Microscopy Laboratory, Bioscience Facilities of the University of British Columbia.

The solid-state electrochemical cell used to perform the experiments at 100% relative humidity, 3 atm pressure, and different temperatures has been described previously.³⁰ Cyclic voltammetry, slow sweep voltammetry, and chronoamperometry were performed using a 50 μm radius Pt microdisk working electrode (Bioanalytical Systems), a Pt gauze counter electrode, and a custom-made dynamic hydrogen electrode (DHE). The reference electrode and the counter electrodes were brought into contact with a piece of Nafion 117 that acted as the backing membrane and worked as the electrolyte. The membrane of interest was held in place at the other side of the backing membrane by the spring-loaded working electrode. All experiments were conducted using a computer driven EG&G PAR model 283 potentiostat.

Prior to an experiment, the surface of the Pt microelectrode was pretreated by potential cycling (30 scans, 1.5 to 0.3 V). A detailed description of the surface pretreatment for the Pt microdisk electrode is given in Ref. ³⁰, together with a description of the operation of the DHE. Potentials are stated relative to the DHE, unless otherwise indicated. Slow sweep voltammograms were performed at 5 mV/s, between 1.5 and 0.3 V. Chronoamperometry was conducted by holding the potential of the microelectrode at 1.2 V for 20 s, before stepping to a potential at which oxygen reduction is diffusion controlled (0.4 V) for 5 s. The current flowing at the microelectrode is given by the modified Cottrell equation

where is the geometric area of the electrode and all the other symbols have their usual meaning. The oxygen diffusion coefficient, and oxygen solubility, in a particular membrane may be obtained from the slope and intercept of an vs. plot provided that, in the time window studied, linear- and spherical-type diffusion both contribute to the overall current.³¹ The permeability of oxygen in each membrane was calculated from the product Electrochemical parameters were determined only for and membranes with sufficiently high IEC to obtain reproducible data.

Three series of polymers with graft chain lengths of 16, 32, and 106 repeating units were synthesized. Polymers were solution-cast to produce flexible and semitransparent films of uniform thickness and good mechanical strength even when hydrated. Membranes were protonated as described in the Experimental section. A summary of the physicochemical properties of these membranes is given in Table I. IECs ranged from 0.45 to 1.85 meq/g, representing SSA mol % values of 2.6 to 13.7. Polymers with higher IECs than those listed were soluble in water, while polymers with lower IECs produced membranes with very low ionic conductivities. Water content ranged from 10 wt % for the lowest IEC membrane to 58% for the highest. Weight percent water contents were generally slightly lower than the volume percents. The averaged proton concentration ranged from 0.40 to 1.01 M in the hydrated membranes. These correspond to λ values of 16 to 36. For comparison, the proton concentration and λ value of Nafion 117 was 1.05 M and 21, respectively.

Table I. 

Properties of membranes.
IEC(meq/g)	SSAcontent(mol %)	Watercontent(wt %)	Watercontent(vol %)	(mol/L)	λ	σ(S/cm)
(16)
1.54	10.7	47	53	0.93	35	0.084
1.15	7.4	35	37	0.78	29	0.031
0.96	6.0	29	32	0.76	27	0.018
0.71	4.3	19	15	0.47	20	0.003
0.45	2.6	10	10	0.40	16	-
(32)
1.76	12.8	50	56	0.97	36	0.100
1.39	8.5	42	48	0.87	32	0.055
1.17	7.6	36	42	0.88	29	0.037
0.98	6.1	28	33	0.90	25	0.013
0.81	4.9	22	23	0.66	20	0.006
(106)
1.85	13.7	51	58	1.01	36	0.098
1.51	10.4	43	47	0.95	31	0.049
1.18	7.6	32	31	0.80	24	0.026
1.13	7.2	24	28	0.81	22	0.013
0.67	3.9	15	17	0.64	16	0.003

Figure 1 plots the water content of membranes as a function of ion exchange capacity. Water content increases with ionic content but is independent of graft chain length. Similarly, proton conductivity of (see Fig. 2a and b) while a strong function of both IEC and λ is independent of graft chain length. This is in contrast to the case of membranes,³² where it has been shown that water content and proton conductivity is strongly correlated to both IEC and graft chain length. An observation from Fig. 2a and b is the exponential growth of conductivity with IEC and λ, in contrast to the near-linear relationship between water content and IEC. These membranes can thus be classified as water-poor³² for which the uptake of additional water by the increase in IEC causes a significant enhancement of proton conductivity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 3 4 5 show TEM micrographs of membranes ultramicrotomed edge-on and possessing different ionic content (mol% SSA). The micrographs indicate the membranes have biphasic morphologies. Isolated ionic domains (dark regions) are evident for membranes possessing the lowest ionic content. However, the TEMs show that increasing the IEC by increasing the number density of graft chains, causes these domains to coalesce into a continuous ionic network. Phase separation is expected since the constituent polymer is comprised of ionic graft chains attached to a nonionic polyacrylonitrile backbone.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Comparison of TEM micrographs of and membranes (see Ref. ³²) shows that the extent of phase separation and connectivity of the ionic domains are similar even though the dielectric constant of polyacrylonitrile is much higher than polystyrene. TEM micrographs, by necessity, are obtained for dehydrated polymers. Upon hydration the hydrophilic domains are expected to swell and become considerably larger than the 10-25 nm size domain observed. In fact, the isolated ionic domains observed in the low ion content samples must coalesce upon swelling in order to support the proton conductivity values observed. The hydrophilic character of the polyacrylonitrile domains has a strong bearing on water uptake. membranes therefore imbibe a much larger quantity of water than their polystyrene-based analogues. This is demonstrated in Fig. 6, which shows water content as a function of ionic content for and membranes comprised of polymers with similar PSSA graft chain length. Water sorption and proton conductivity values for other graft poly(styrenesulfonic acid) membranes are summarized in Table II and plotted in Fig. 6 and 7. The data show that within a series of membranes water content increases with IEC. The upper limit of IEC for and Nafion-type membranes is determined by their dissolution in water. The ETFE- and PVDF-based membranes employed in this comparative study are prepared from preformed hydrophobic films. They can therefore be fabricated with much higher IECs, and they imbibe larger quantities of water without dissolution.

Zoom In Zoom Out Reset image size

Table II. 

Conductivity and water content for various PSSA grafted membranes. Nafion 117 included for comparison.
IEC(meq/g)	Watercontent(wt%)	Watercontent(vol%)	(mol/L)	λ	σ(S/cm)	Ref.
(29)
1.68	34	37	0.92	19	0.15	32
1.51	32	38	1.01	18	0.09
1.29	21	14	1.00	12	0.04
1.01	15	18	0.86	10	0.02
0.81	11	10	0.71	8	0.01
0.62	9	8	0.58	8	-
3.27	61	74	1.32	28	0.30	23, 25
2.56	49	66	1.42	23	0.22
2.45	47	64	1.76	22	0.20
2.13	42	53	1.52	22	0.19
3.06	61	82	1.55	28	0.32	This work
2.69	51	72	1.89	21	0.25
2.24	44	68	1.87	19	0.20
1.82	38	53	1.79	19	0.15
1.37	29	50	1.52	17	0.10
Nafion 117 (E)a
0.91	26	39	1.05	21	0.10	23
a Expanded form.

Zoom In Zoom Out Reset image size

Plots of conductivity vs. IEC and (λ) are shown in Fig. 7. They show that the conductivity within a series of membranes is dependent on the ionic content and water content. An explanation of the high conductivity of the membranes has been previously reported, and is related to the high in these membranes.²⁵ The same explanation applies to membranes. The plots illustrate that with these radiation-grafted membranes, conductivity increases with increasing λ, indicating that additional water assists proton conduction. Two anomalies deserve explanation. The first is that membranes exhibit lower conductivities than ETFE and PVDF radiation-grafted membranes even though they possess similar values of λ. This is due to the lower IECs of membranes which limit their water uptake and result in lower proton concentrations. Calculated values of are presented in Tables I and II. Second, even though the series of membranes possess similar IECs and but much larger λ values than the series, proton conduction through membranes is smaller for a given ion content. This discrepancy is assumed to be due to water that swells the partially hydrophilic PAN domains in addition to sulfonic acid domains, and that this water does not assist proton conduction to the same degree. This is discussed in more detail in the next section in conjunction with the electrochemical characterization.

Chronoamperometry was used to determine ORR mass transport properties in the membranes at the Pt/membranes interface. Oxygen diffusion coefficients and oxygen solubilities were obtained by linear regression analysis of vs. plots. Table III summarizes the data. All data were obtained using the same instrumentation and under identical conditions (323 K, 3 atm pressure, 100% RH).

Table III. 

Oxygen mass transport parameters for various membranes at 323 K and 3 atm pressure.
IEC(meq/g)	Watercontent(vol %)	Solubility	Diffusioncoefficient	Permeability	Ref.
1.85	58	1.8	15.6	28	This work
1.76	56	2.3	17.8	41
1.54	53	2.0	11.2	22
1.51	47	2.1	7.3	15
1.39	48	1.8	7.0	12
1.15	37	1.9	3.7	7
1.68	34	9.6	1.7	16	This work
1.55	33	9.2	2.6	24
1.38	29	8.3	4.0	33
1.31	22	7.9	1.7	13
1.24	21	10.3	1.8	18
1.04	9	14.1	0.8	11
3.27	74	3.3	16.6	54	24
2.56	66	4.1	12.2	50
2.45	64	4.4	9.3	40
2.13	53	3.8	8.3	32
3.06	82	3.4	17.9	61	This work
2.69	72	2.9	17.2	49
2.24	68	2.9	14.4	43
1.82	53	3.7	9.2	34
1.37	50	3.2	4.7	15
Nafion 117 (E)a
0.91	39	5.5	9.3	51	24
a Expanded form.

Figure 8 plots the effect of IEC on oxygen mass-transport properties. diffusion coefficients increase with increasing IEC because water content increases with IEC and oxygen diffusion occurs largely through aqueous domains.³¹ In contrast, oxygen solubility remains relatively constant within a series. Oxygen solubility is favored in hydrophobic zones and surprisingly is not affected within a series to a great extent by variations in IEC and water content. As a consequence, the increase in oxygen permeability with increasing IEC follows the same trend as the diffusion coefficient.

Zoom In Zoom Out Reset image size

Other pertinent observations include the high permeability of oxygen in Nafion compared to other membranes with similar IEC, and the relatively low diffusion coefficients but high oxygen solubility associated with the membranes. More insight into these results is obtained from comparisons of oxygen mass-transport properties plotted as a function of water content (Fig. 9). Figure 9a clearly demonstrates that diffusion coefficients increase with increasing water content, and explains why and membranes allow higher diffusion coefficients than Nafion 117. The membranes support a higher diffusion coefficient for a given water content compared to and This is because the λ values are larger for membranes and therefore, the water within the membrane is less coordinated to ionic groups. In the case of membranes, the water contents and λ values are much lower than in and consequently diffusion coefficients are much lower.

Zoom In Zoom Out Reset image size

In contrast to the diffusion coefficients, oxygen solubility in membranes is much higher than in and membranes because the former are relatively hydrophobic and their water contents lower. The membranes solubilize less oxygen than the and membranes for a given water content because the polyacrylonitrile backbone is more hydrophilic than ETFE or PVDF. The high solubility of oxygen in Nafion is due to the nonpolar fluorocarbon matrix.³¹ The increase in oxygen permeability as a function of increasing water content is largely dominated by the increase in diffusion coefficient. The radiation-grafted membranes are highly permeable because of their large IEC and correspondingly high water content. Nafion 117 exhibits an unusually large permeability to oxygen given its modest water content. This is because the diffusion coefficient is higher than expected based on the trends of other membranes (Fig. 9b) and because the oxygen solubility is significantly higher. This is likely a direct consequence of phase separation, which provides a continuous hydrophilic network for diffusion and hydrophobic domains for oxygen solubility.

Since mass-transport limiting currents for the ORR are directly related to oxygen permeability, one might conclude that those materials possessing high oxygen permeability would be more suitable for fuel cell operation.²⁴ However, for purposes of preventing crossover of reactant gases, low permeability may be required. Notwithstanding issues associated with membrane stability and processability, it would appear worthwhile to investigate mulitilayered PEMs or graded composition PEMs that are highly permeable to oxygen at, or near, the catalyst/membrane interface, and less permeable in the bulk of the membrane.