Article pubs.acs.org/ac
Analysis of the Hydration Process and Rotational Dynamics of Water in a Nafion Membrane Studied by 1H NMR Spectroscopy Chihiro Wakai, Takafumi Shimoaka, and Takeshi Hasegawa* Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan S Supporting Information *
ABSTRACT: 1H NMR spectroscopy is employed to reveal the hydration process of a Nafion membrane by measuring both the chemical shift and the spin−lattice relaxation time. In a former study, the hydration process was suggested to comprise two steps: the molecular adsorption of water on the sulfonic acid groups and wetting with liquid water. The present study has revealed the first step can further be divided into two steps. By introducing a new experimental technique, the quantitatively reliable NMR measurements of protons (1H) of water involved in the polymer membrane are realized. In addition, a new analytical procedure is developed using a reciprocal concentration on a saturation−adsorption model, and the hydration is clearly revealed to have three individual steps. Both the chemical shift and the relaxation time plots against the reciprocal concentration exhibit three linear parts with apparently different slopes. Of great interest is that the initial hydration is divided into two stages: the first hydration is a very strong adsorption of water probably on the hydroxyl group of the sulfonic acid group, and the second one is a relatively weak adsorption on another site of the sulfonic acid group. The third hydration is readily assigned to excess bulk (liquid-like) water as expected. These adsorption processes are readily correlated with the rotational motion of water by converting the spin−lattice relaxation time to the rotational correlation time.
A
not only the hydrophilic property but the locally oriented structure. IR spectroscopic analysis also reveals the hydration process. Iwamoto et al.1 reports that a drying process of a wet Nafion membrane can be pursued by IR spectroscopy, and the hydration is separated into two cases: condensed liquid water and strongly bound molecular water to the sulfonic acid. The dissociation equilibrium of the sulfonic acid group is also revealed by some key IR bands. In this manner, both water and the sulfonic acid group are readily analyzed during the drying process. On the other hand, 1H NMR is a powerful technique to pursue both quantity and motion of water involved in a Nafion membrane. In a recent paper, Zhao et al.5 reports interesting analytical results of equilibrium swelling water using 1H NMR. They reveal that the adsorbing water can be divided into two types. Type 1 is the strongly adsorbing water on the sulfonic acid groups to form a water shell, while type 2 is the weakly adsorbing water molecules on the water shell. When the number of adsorbed water molecules per one sulfonic acid group, λ, is less than four, the water adsorbs in the manner of type 1, whereas the additional water adsorbs in the manner of type 2 when λ is more than four. The type 2 water is thus concluded to have a character of liquid water, which is quite reasonable along with the IR results.
Nafion membrane is a key device in a polymer electrolyte membrane fuel cell (PEMFC). The membrane works as a separator of oxidizing and reducing agents, but at the same time it transports proton between the agents. Nafion is a copolymer of sulfonated tetrafluoroethylene and perfluoroalkyl ether (Chart 1), and the property of proton transportation is due
Chart 1. Primary Chemical Structure of Nafiona
a
For detail, please refer to the text.
to the ionomer structure. The performance of the membrane is known to be degraded when the membrane is swollen by absorbing water and methanol. Since the proton transfer occurs along with the migration of water across the membrane, the hydration process of Nafion has long been attracting much interest.1−8 The molecular structure of a Nafion thin membrane has recently been discussed by infrared (IR) spectroscopic technique1,2 that the sulfonic acid group has a specific association and orientation in the membrane. Since the hydrophobic part that occupies more than 90% of the membrane exhibits nearly no molecular orientation, the role of the sulfonic acid groups should be very important in terms of © 2013 American Chemical Society
Received: June 4, 2013 Accepted: July 11, 2013 Published: July 11, 2013 7581
dx.doi.org/10.1021/ac401653v | Anal. Chem. 2013, 85, 7581−7587
Analytical Chemistry
Article
the mixing time. The FID signals were accumulated 8 times using 30 different mixing times, which took 5 min. Sampling Technique of a Thin Membrane for NMR. 1H NMR spectra of a Nafion membrane were measured as a function of time under the humid condition at 30 °C with the experimental setup as illustrated in Figure 1a. In the NMR tube
In the present study, a new sampling technique has been developed for obtaining quantitatively reliable NMR spectra of a thin wavy membrane sample as is. The reproducibility of the peak position and area were both largely improved, and the time-course measurements during the hydration of a Nafion membrane were readily performed. In addition, a new theoretical model has also been developed using the reciprocal concentration. The new technique predicted that the hydration process should be represented by a linear function until the water-receiving site has been saturated. In fact, the peak area and the relaxation time both exhibited consistently three apparently distinguishable linear regions against the reciprocal concentration. Of note is that the first two regions both correspond to the type 1 water. In other words, the initial hydrating water that strongly binds to the sulfonic acid group should be understood in two adsorption steps. The spin−lattice relaxation analysis also reveals the rotational dynamics of water involved in the Nafion membrane. The two kinds of water in the hydration step exhibit much longer periods of time than the third water, which means that initial two water species are discriminated from each other also in terms of the rotational dynamics. The restriction supports that both types of water molecules are bound water on the sulfonic acid group, which are apparently recognized to be different from the liquid-like water.
■
EXPERIMENTAL SECTION AND METHODS Materials. The sample membrane was a Sigma-Aldrich (St. Louis, MO, U.S.A.) Nafion NRE-212 membrane (Chart 1). The index, m, is calculated to be 6.5 based on the data provided by the supplier, while the other indices cannot be determined because the molecular weight of Nafion is not available. The membrane was cut into a piece with a size of 2 mm × 50 mm. The sample membrane was boiled in a 3% H2O2 aqueous solution at 100 °C for 1 h, followed by another boiling in 1 M H2SO4 aqueous solution at 100 °C for 1 h, and rinsed by pure water. Since the spin−lattice relaxation time was observed more than 200 ms, the paramagnetic impurities such as Mg and Fe were recognized to be removed from the sample.1 The washed sample membrane was dried in an oven at 40 °C for 12 h. In addition, the membrane was adequately dried by flowing argon gas for two hours in the NMR tube. The water used in the study was obtained by a Millipore (Molsheim, France) Elix UV-3 pure-water generator and a Yamato (Tokyo, Japan) Autopure WT100U water purifier, which is a compatible model with Milli-Q. The water exhibited an electric resistivity higher than 18.2 MΩ cm. 1 H NMR Measurements. The one-dimensional 1H NMR spectra and spin−lattice relaxation time, T1, were measured for 1 H (proton) involved in the hydrated Nafion membrane by using a JEOL (Tokyo, Japan) ECA600 NMR spectrometer. Pure heavy water was used as the external reference of the chemical shift, and the standard signal was set to 4.72 ppm. The 1 H NMR spectra and T1 were both measured at 30 °C in a dry condition at t = 0 h. After setting a water vessel (3 mm o.d.) filled with pure water, the 1H spectra and T1 were measured every 15 min. The time for measuring the free induction decay (FID) signal was 1 min, and it was accumulated 16 times to improve the signal-to-noise ratio of the one-dimensional NMR spectra. To measure T1, the inversion−recovery method was used. The pulse sequence was π−tmix−π/2, where π and π/2 are the nonselective 180° and 90° pulses, respectively, and tmix is
Figure 1. Schematics of the sample measurements in a NMR tube: (a) membrane without a weight; (b) membrane spread with a weight.
(3 mm o.d.), the Nafion membrane is placed as is at the bottom, and a water vessel is hung above the sample. With this setup, the ambient air in the tube is highly humid, which is making the membrane hydrated with time. The time-course NMR measurements were performed during the hydration, and the representative two spectra are presented in Figure 2, parts a and b. The most significant peak at about 7 ppm comes from water, but the peak is not a single-component one, and the location is
Figure 2. 1H NMR spectra of water involved in the Nafion membrane. The spectra of 1 and 2 present poor reproducibility. The improved spectrum by using the weight-hanging technique is presented at part c. 7582
dx.doi.org/10.1021/ac401653v | Anal. Chem. 2013, 85, 7581−7587
Analytical Chemistry
Article
To do that, a glass-ring weight was put on the bottom of the membrane as illustrated in Figure 1b. With this simple technique, the NMR spectra measurements were drastically improved as presented at Figure 2c. The satellite artifact peaks are largely removed, and the main single peak has become very strong with a good reproducibility. IR Attenuated Total Reflection Measurements. The water remained in the membrane can be analyzed by using 1H NMR. When the amount of water is very minute, however, the NMR peak becomes very broad and the peak area is difficult to analyze. To check the state of wetness of the membrane, therefore, IR spectroscopy was employed. For the temperaturedependent IR measurements, the attenuated total reflection (ATR) technique was used. The spectra were recorded by a Jasco (Tokyo, Japan) model 6100 FT-IR equipped with a custom-made heatable ATR accessory, model ATR Pro610P-S, which can elevate the sample temperature up to 120 °C. Unpolarized measurements were performed using a TGS detector, and the modulation frequency was 10 kHz. The IR spectra are presented in Supporting Information Figure S-1. The water-related band is found at about 3400 cm−1, and it disappears rapidly over 60 °C, and it is nearly completely lost at 120 °C. This was confirmed by 1H NMR measurements at 120 °C. As presented in Supporting Information Figure S-2, the curve converges to 11.1 ppm at the thoroughly dried state ([H]0/[H]t = 1 as mentioned later), which can be used as a marker of a dried Nafion membrane.
not reproduced at all. In addition, the tail structure on the lowfield (high-δ) side of the peak is not reproduced either, and new peaks near 8 ppm are found in the spectrum of Figure 2b. The poor reproducibility can be attributed to the nonuniform shape of the thin wavy membrane as discussed below. Since a thin polymer membrane with no support is flabby, it has wavy and wrinkled local shapes. The shape with the complicated curvatures should cause an inhomogeneous magnetic field in the sample. The magnetic field in the sample, HS, is correlated with the shape factor, α, and the magnetic susceptibility, χS, as follows:8−10 ⎧ ⎛4 ⎞ ⎫ HS = H0⎨1 + ⎜ π − α⎟χS ⎬ ⎝ ⎠ ⎭ ⎩ 3
(1)
H0 stands for the magnetic field in vacuum. The sample shape is reflected in the shape factor. For example, a cylindrical shape takes the shape factor of 0 and 2π for the parallel and perpendicular configurations to the external magnetic field, respectively. For the spherical shape, α is 4π/3. In the present study, heavy water (deuterium oxide) in a normal NMR tube (α = 0) was used as the external reference. The magnetic field in the reference, HR, is then described as follows: ⎛ ⎞ 4 HR = H0⎜1 + πχR ⎟ ⎝ ⎠ 3
(2)
Here, χR is the magnetic susceptibility of heavy water. With these parameters, the chemical shift, δ, can be described as follows by considering the Larmor equation (ω = γμ0H) and the two good working approximations: χR ≪ 1 and σR ≪ 1. ω − ωR δ≡ S × 106 ωR =
HS(1 − σS) − HR (1 − σR ) × 106 HR (1 − σR )
≈
{( σ
R
− σS) −
4 π (χ − χS ) − αχS 3 R
} × 10
■
RESULTS AND DISCUSSION H NMR spectra measured at t = 0, 0.5, 1, 2, 5, and 20 h are shown in Figure 3. At t = 0, a very broad peak appears at about 1
6
(3)
Here, ωR and ωS are the Larmor frequencies and σR and σS are the shielding constants of the reference and sample, respectively. In the case of the cylindrical sample like an NMR tube aligned parallel to the external magnetic field, the observed chemical shift becomes as follows because of α = 0. δcylindrical =
{( σ
R
− σS) +
4 π (χ − χS ) 3 R
} × 10
6
(4)
Figure 3. 1H NMR spectra measured at t = (a) 0, (b) 0.5, (c) 1, (d) 2, (e) 5, and (f) 20 h under the humid atmosphere.
This holds for the “external reference” method normally employed for NMR measurements of solutions. The magnetic susceptibilities for usual materials have similar values, and they are in the range of the order of 10−6. Therefore, the second term of eq 4 is negligible, which enables us to easily discuss the NMR peaks using the shielding constants only. For the NMR analysis of 1H involved in a thin membrane, on the other hand, α varies a lot depending on the local shape of the membrane, which cannot be ignored. In other words, artifact peaks are thus generated by the wavy shape of the sample. The magnetic susceptibility is primarily governed by the diamagnetic term, and χS is thus negative in the present case. In fact, the artifact peaks appear on the low-field side of the water peak as recognized in Figure 2. Therefore, the artifact peaks would get back to the main peak by removing the membrane wrinkles away.
10.2 ppm. The large width of the peak tells us that the motion of proton in analysis should largely be restricted, and nearly no proton exchange is recognized. This means that the broad peak is assigned to the proton of the sulfonic acid (−SO3H) group with very few water molecules in Nafion. In the sulfonic acid group, the electron of the hydrogen atom is strongly drawn by the SO3 group; the broad peak appears at a high chemical shift. With increasing the water adsorption time, this peak shifts to the higher-field (low-δ) side because the proton receives electron from the adsorbed water via the hydrogen bonding, and the bandwidth becomes sharper apparently. The sharpness is also a result of the proton exchange between the sulfonic acid group and water, in which the protons cannot be distinguished from each other. In this manner, the sharp peak is assigned to 7583
dx.doi.org/10.1021/ac401653v | Anal. Chem. 2013, 85, 7581−7587
Analytical Chemistry
Article
the averaged peak of sulfonic acid group and adsorbed water molecules. The minor peak at 3.2 ppm is considered to be due to an impurity that remained, since the peak does not shift in the time course. In Figure 4, the position of the proton peak is
Figure 5. Time dependence of the normalized proton concentration due to the −SO3H group and water molecules involved in the Nafion membrane.
the observed result is obtained as a weighted average of the independent sites. On the adsorption model, one of the independent sites, R−SO3H, is a constant in quantity, while only the adsorbing water molecules increase in quantity until the water-receiving sites are fully occupied. The individual proton quantities of the R−SO3H and H2O species (C1 and C2, respectively) are thus necessary to make quantitative discussion. As mentioned, C1 remains constant, [H]0, while C2 increases with time, which can be expressed as
Figure 4. Time dependence of the chemical shift of a proton of the −SO3H group and water molecules involved in the Nafion membrane.
plotted. At a very initial stage, the position gradually changes in about 1 h. After that, the position quickly moves down and the hydration process looks like a continuous change. To discuss the change quantitatively, equilibrium is taken into account. Before discussing details on this continuous hydration model, the increase of the total proton concentration, [H]t, normalized by the initial concentration, [H]0, is plotted against time in Figure 5. The total concentration was obtained by integrating the NMR peak in the range of 4−14 ppm. The monotonously increasing curve is found, but on closer inspection, the curve looks to consist of three regions of [H]t/[H]0 = 2.0−3.6, 3.6− 6.3, and 6.3−7.8 with different slopes, which roughly indicates that there are about three stages for elucidating the hydration. To discuss the hydration process in more detail, the chemical shift (δ) is discussed with an equilibrium model. In the thoroughly dried condition, at the beginning of the hydration process (t = 0), the proton concentration in the Nafion membrane, [H]0, is equal to the concentration of the electronically neutral R−SO3H group, which is denoted as C1. C1 ≡ [H]0 = [R−SO3H]
C2 ≡ [H]adsorbed = [H]t − [H]0
Here, [H]t is the total quantity of proton found in the membrane at t = t. By using the quantity parameters, a physical observable, y, is the weighted average of the two independent parameters, y1 and y2: y = y1
C1 C2 + y2 C1 + C2 C1 + C2
= y1
[H]0 [H]t − [H]0 + y2 [H]t [H]t
= y2 + (y1 − y2 )
[H]0 [H]t
(8)
In this manner, y is linearly correlated with the reciprocal concentration, [H]0/[H]t. With respect to eq 8, y2 is easily obtained as the intercept on the ordinate. On the other hand, the slope, y1 − y2, is also a constant, since only C2 changes with time when the adsorption model works correctly. Then, the results in Figure 5 are replotted against the reciprocal concentration in Figure 6. As expected, three linear parts appear apparently in the new curve except region 2′. The results indicate that the adsorption model on a saturated surface works appropriately, and there are three different adsorption stages during the hydration. When the next adsorption begins after the first adsorption reaches the saturation, however, the formulation changes. Let us consider that the water molecules occupy the adsorption
(5)
When the membrane receives water, a part of the sulfonic acid groups are hydrated to have the complex of R−SO3H···H 2O ⇌ R−SO3−·H3O+
(7)
(6)
The equilibrium means that a proton is shared by the two species via a very fast exchange. The time constant of the exchange is known to be about 10−10 s.11 Since this time constant is much shorter than that of the NMR measurements (∼100 s), the proton cannot be attributed to one of the species, and as a result a single NMR peak is generated. In other words, 7584
dx.doi.org/10.1021/ac401653v | Anal. Chem. 2013, 85, 7581−7587
Analytical Chemistry
Article
Table 1. Analytical Results of δ, R1, and τc in the Three Regions Obtained by Extrapolating Each Linear Part in Figures 6 and 7 [H]0/[H]t region region region region
sites of the R−SO3H groups with a molar ratio, n, on the first stage. In this case, the quantity of the saturated sites becomes (9)
With this notation, the quantity of the third adsorbate at t = t can be expressed as C3 ≡ [H]additional = ([H]t − [H]0 ) − n[H]0 = [H]t − (n + 1)[H]0
(10)
With this parameter, the physical observable, y, of the threecomponent system is rewritten as C1 C2 + y2 C1 + C2 + C3 C1 + C2 + C3 C3 + y3 C1 + C2 + C3
y = y1
= y1
[H]0 n[H]0 [H]t − (n + 1)[H]0 + y2 + y3 [H]t [H]t [H]t
= y3 + {y1 + ny2 − (n + 1)y3 }
[H]0 [H]t
R1/s−1
9.52 6.73
3.54 8.76
4.61
0.64
τc/ns 220 8.8 0.009
example, the chemical shift of the adsorbing species is obtained to be 4.61 ppm by the extrapolation of the linear part, which is very close to that of pure liquid water (4.7 ppm).12 Of course, the analytical error may be large because of the extrapolation, which may reach up to ±20%. Regardless, the order of magnitude can readily be discussed with the analytical results. In this manner, the additional species in region 3 has readily been assigned to be the excess bulk water, (H2O)n. Region 2′ is an exceptional part, since the linearity is unclear, which suggests that two kinds of adsorption (regions 2 and 3) go parallel. Since the analytical model is that an adsorption occurs on a saturated system, the discussion of the unclear linear part should be avoided. The rest two regions are of great interest. Judging from the chemical shift, the two regions should be related to “molecular water” directly adsorbed on the sulfonic acid groups. In our initial speculation before the analysis, the direct adsorption of water on the groups should simply be continued until the water-receiving groups are fully occupied. Unexpectedly, however, the water adsorption process is suggested to comprise two processes. On the definition of the reciprocal concentration, [H]0/[H]t = 1 corresponds to the dried condition (t = 0). At the very initial stage of the hydration, in region 1, the chemical shift is obtained to be 9.52 ppm, which suggests that the proton exchange is relatively slow. The high chemical shift is consistent with the fairly large peak width as found in Figure 1a. As a result, the adsorption on the sulfonic acid groups is concluded to be quite strong. When this adsorption reaches the saturation, at [H]0/[H]t = 0.43, [H]t/[H]0 becomes 2.33, which means that the number of the additional protons are 1.33 (= 2.33 − 1) per one sulfonic acid group. Since the additional protons are provided by the adsorbing water, the number of water is obtained as 0.67 (= 1.33/2) per a sulfonic acid group. In other words, about 70% of sulfonic acid groups work as the waterreceiving sites for molecular water, if one sulfonic acid group receives one water molecule. The water receiving site should be attributed to the O−H group of the sulfonic acid group. Immediately after the first saturation, the second adsorption begins in region 2 that finishes at [H]0/[H]t = 0.25, which gives the chemical shift of 6.73 ppm (Table 1). Since [H]t/[H]0 is 4.00, the total number of the water molecules after the second adsorption is calculated to be 1.50 per one sulfonic acid group. Roughly speaking, in the second adsorption process, the number of adsorbed water molecules has become doubled. In other words, one sulfonic acid group may receive more than two water molecules. Judging from the peak position (6.73 ppm), the second adsorption species should also be attributed to water strongly bound to the sulfonic acid group. At the moment, details about the receiving site for the second adsorption water are not clear, but the present analysis strongly suggests that the hydration process comprises two steps that occur successively.
Figure 6. Data of Figure 5 are replotted between the chemical shift and the reciprocal concentration of proton.
C2 ≡ n[H]0
1: 0.43 < 2: 0.25−0.43 2′: 0.19−0.25 3:
