High Field MAS NMR and Conductivity Study of the Superionic

Jun 13, 2014 - Jin Jung Kweon†, Riqiang Fu*†, Eden Steven†, Cheol Eui Lee‡, and .... E. Lee , J. H. Han , D. Kim , G. W. Jeon , Cheol Eui Lee ...
1 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/JPCC

High Field MAS NMR and Conductivity Study of the Superionic Conductor LiH2PO4: Critical Role of Physisorbed Water in Its Protonic Conductivity Jin Jung Kweon,† Riqiang Fu,*,† Eden Steven,† Cheol Eui Lee,‡ and Naresh S. Dalal*,†,§ †

National High Magnetic Field Laboratory, Florida State University, 1800 E. Paul Dirac Drive, Tallahassee, Florida 32310, United States ‡ Department of Physics and Institute for Nano Science, Korea University, Seoul 136-713, South Korea § Department of Chemistry and Biochemistry, Florida State University, Tallahassee, Florida 32306, United States ABSTRACT: LiH2PO4 (LDP) is a favored candidate for hydrogen fuel cells, but the mechanism of its high protonic conductivity remains unclear. A complicating factor has been the lack of resolution in the reported proton NMR spectra. We now report multinuclear magic angle spinning NMR in LDP at magnetic fields up to 21.2 T. Well-resolved 1H NMR spectra are observed that are assignable to protons in the short and long O−H···O hydrogen bonds and a peak to physisorbed H2O. The position and intensity for the H2O peak depend on the H2O content, implying fast exchange between the adsorbed H2O and the O−H···O protons. 31P and 7Li NMR spectra and spin−lattice relaxation measurements showed that the proton hopping/exchange processes involve concerted hindered rotational fluctuations of the phosphate groups. Conductivity data from adsorbed H2O-controlled samples clearly suggest that the mechanism of LDP’s protonic conductivity is dominantly the exchange (and hopping) of the adsorbed H2O protons with the short O−H···O hydrogen bonds, in contrast to an earlier model that ascribed it to intermolecular hopping of O−H···O protons. The new findings enable us to modulate LDP’s protonic conductivity by several orders of magnitude via controlling physisorbed water. of 2.677 and 2.561 Å connecting the tetrahedral PO4 groups.4,10 In several earlier studies of LDP, fast interbond proton hopping was suggested to be responsible for the exceptionally high protonic conductivitythis conclusion being based on the correlation between temperature-dependent 1H spin−lattice relaxation and conductivity measurements.8,11,12 However, in the similar compound CDP, which also contains two different H-bonding environments, the interbond proton hopping was found to be very slow, together with low protonic conductivity (10−9−10−10 Ω−1 cm−1).6,13 More importantly, LDP is known to be hygroscopic,7 but whether or not the adsorbed H2O plays any role in LDP’s protonic conductivity mechanism was not considered. This is important because earlier studies of several AH2PO4 compounds (A = Li, K, and Cs) had indicated the role of lattice dehydration or humidity in the conductivity mechanism. For example, the ionic conductivity of CDP and its composite with silica was found to increase, reversibly, above 230 °C in humid conditions, the conductivity being influenced by silica’s hydrophilicity.14 The acceleration of the conductivity could be induced by point defects generated in the ionic salt

1. INTRODUCTION Understanding the mechanism of proton conduction in hydrogen-bonded solids is of considerable interest owing to their potential application as proton conduction electrolytes in fuel cells. Despite focused effort the basic mechanism remains unclear.1−3 The AH2PO4-type (A = Li+, NH4+, K+, Rb+, Cs+, Tl+) hydrogen-bonded lattices show various physical properties depending on the ionic radius of the cation A and the hydrogen bond lengths.4 For instance, at room temperature (RT) the protonic conductivity was measured to be 10−10−10−11 Ω−1 cm−1 for KH2PO4 (KDP)5 and 10−9−10−10 Ω−1 cm−1 for CsH2PO4 (CDP).6 In the so-called superprotonic phase transition at 230 °C, an increase of protonic conductivity by up to 4 orders of magnitude has been observed for CDP, resulting from rapid dynamic motion between two different Hbonding environments and fast rotation of the phosphate ions.2 On the other hand, LiH2PO4 (LDP) exhibits an exceptionally high protonic conductivity (i.e., 10−3−10−5 Ω−1 cm−1) even at RT,7−9 but the underlying conduction mechanism has not been fully understood, thus suggesting an attractive research area. A good protonic conductor generally involves O−H···O hydrogen bonds longer than 2.60 Å.3 The crystal structure of LDP determined by X-ray and neutron diffraction shows two types of O−H···O hydrogen bonds with different bond lengths © 2014 American Chemical Society

Received: February 12, 2014 Revised: June 1, 2014 Published: June 13, 2014 13387

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393

The Journal of Physical Chemistry C

Article

(CDP) near the surface between the ionic salt and silica.14 Similarly, a positive effect of high temperature steam on protonic conductivity of CDP/SiP2O7-based composite electrolytes has been reported.15 Observation of increase in ionic conductivity in RbH2PO4 under high humidity to suppress its lattice dehydration (i.e., loss of hydrogens and conversion to Rb2H2P2O7) has also been studied.16 The mechanism of protonic conductivity in KDP has been studied in detail.17 The mobile species were thought to result from a 3-fold rotation of H2PO4− units about any axis leading to H···O···H···O units. The H···O···H bridge is dissociated under applied voltage to lead to H+ and OH− moieties. But the role of surface adsorbed water was not mentioned. We note, in addition, that the role of physisorbed H2O in conductivity of conducting polymers such as polyaniline and silica glasses has been mentioned.18−21 Highperformance CDP-based fuel cells have been reported and humidity was found to be helpful for their stable performance.22 Finally, in LDP itself, the role of humidity has been noted,9 but surprisingly, the conductivity was reported to decrease with increase in dampness, in contrast to CDP. In addition, the earlier NMR studies on LDP were carried out at the low frequency of 200 MHz for 1H, where spectral resolution was insufficient for assignment of peaks to specific O−H···O protons,11,12 adding to the impetus for the present undertaking. In the present study we utilized the much higher resolution afforded by the 300, 600, and 900 MHz solid-state NMR instrumentation, available at the National High Magnetic Field Laboratory (NHMFL). Indeed, using magic angle spinning (MAS) at these high frequencies (high fields), peaks from the different O−H···O protons were well resolved and easily assignable to the “long” and “short” hydrogen-bonded protons. An additional narrow peak was observed from adsorbed H2O that actually dominated the spectra. Parallel 1H and 31P NMR and protonic conductivity studies on samples containing (relatively) known amounts of adsorbed H2O clearly suggest that the exchange between the adsorbed H2O and O−H···O protons and interbond hopping of protons, in concert with the rotational fluctuation of the phosphate moieties is the dominant mechanism of LDP’s protonic conductivity. This finding enabled us to control the protonic conductivity of LDP for the first time, by a careful control of physisorbed H2O, suggesting an additional important factor in the utility of LDP as a superionic material.

show strong enough resonances from the long T1 components. Adamantane (1.63 ppm, relative to TMS) was used as the chemical shift reference. A cross-polarization (CP) based heteronuclear 31P−1H correlation spectrum with a spinning speed of 12.5 kHz was recorded at 600 MHz using a CP contact time of 50 μs and a recycle delay of 5 s with a spin-lock field of 35 kHz. Similarly, direct polarization (DP) was used to record single-scan 31P and 7Li spectra at 900 MHz using 13.5 kHz spinning speed, again acquired right after the sample was first inserted into the magnet. The chemical shifts for 31P and 7 Li spectra were calculated from the 1H chemical shift based on their relative gyromagnetic ratios. A short recycle delay of 0.01 s was used in the saturation-recovery method for T1 measurements of 31P and 7Li. Recrystallized LDP powder was fabricated into a disk sample (12 mm diameter, 1 mm thickness) by uniaxially pressing in a cylindrical die at 4000 kg/cm2 in ambient atmosphere. DC and AC electrical conductivity measurements as a function of vacuum pumping time were made on the pellet samples in an Ar-filled glovebox by a simple two-probe technique using an Extech MN15A Digital Mini multimeter for DC measurement and a Gamry Reference 600 potentiostat/galvanostat for AC measurement. Along with AC electrical conductivity measurement at 10 Hz, the change in the amount of adsorbed H2O was determined from the mass change of LDP the pellet as a function of pumping time. To measure humidity-dependent conductivity, three pellet samples were made using different pressures of 2000, 4000, and 6000 kg/cm2. Simultaneous AC conductivity measurements were made on the three pellet samples at 20 Hz as a function of relative humidity at 40 °C inside a humidity-controlled chamber using three different lockin amplifiers (Stanford Research Systems, Model SR830 DSP).26

3. RESULTS AND DISCUSSION 3.1. Peak Assignment. Figure 1a shows a comparison of the low-field, low-resolution and high-field, high-resolution spectra obtained at RT using a recrystallized powder sample under the same humidity. A narrow peak at ∼8 ppm was observed at 300 MHz, while three resonances were observed at

2. EXPERIMENTS High-resolution MAS NMR measurements were made using polycrystalline LDP powder (99.99%), purchased from Aldrich, and purified by recrystallization in double processed purified water at 30 °C, following earlier reports.23−25 1H MAS NMR line shape and spin−lattice relaxation time (T1) measurements were made using 300, 600, and 900 NMR spectrometers on samples with a systematic control of physisorbed water by pumping. On the 300 and 600 MHz NMR spectrometers, 4 mm Bruker CPMAS double-resonance NMR probes and a 3.2 mm homemade Low-E MAS NMR probe were used with a spinning rate of 13.5 kHz. A locally designed and fabricated 3.2 mm CPMAS double-resonance NMR probe was used on the 900 MHz spectrometer. In order to quantify relative signal intensity between the short and very long T1 components, single-scan 1H NMR spectra were acquired right after the sample was first inserted into the magnet using the spin−echo pulse sequence to remove background 1H signal from the probe. Multiscan spectra with short relaxation delay did not

Figure 1. (a) 1H spectra at 300 MHz using 7 kHz spinning speed (top), 600 MHz using 13.5 kHz (middle), and 900 MHz using 13.5 kHz (bottom) of the recrystallized LDP powder without vacuum pumping, showing resolution enhancement at high fields. The small peak with smaller chemical shift than the water exchange peak acquired at 600 MHz is background proton signal from a 3.2 mm homemade Low-E MAS NMR probe. (b) 1H spectra at 600 MHz with a systematic control of adsorbed water by vacuum-pumping time. 13388

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393

The Journal of Physical Chemistry C

Article

600 and 900 MHz: one at ∼8 ppm and two broad peaks at 12.4 and 8.4 ppm. Figure 1b shows the 1H spectra with carefully H2O-controlled samples, clearly indicating three well-resolved resonances. It is known that there is a direct correlation between the interoxygen (O−H···O) distance and the 1H NMR chemical shift:27,28 the resonance at a smaller chemical shift is attributed to the protons in the long hydrogen bonds, while the peak at a larger chemical shift to protons in short hydrogen bonds. Therefore, the two broad peaks at 12.4 and 8.4 ppm can be assigned to the two protons in two different hydrogen bonds: H1 with the longer O−H···O length of 2.677 Å and H2 of the shorter length of 2.561 Å.9 Their large line widths imply that the motion involving these protons is rather slow. The third, narrow peak, on the other hand, corresponds to highly dynamic protons. As shown in Figure 2, both broad peaks are

peak, and thus it is hard to follow its exchange under our conditions at any NMR frequency used. Nevertheless, it is clear that under higher water content the proton exchange is dominantly with protons from the short hydrogen bond lengths. 3.2. Role of Water in LDP System. A clue to the lack of role of interbond hopping was suggested by the widths of the proton peaks: the widths of the peaks from both sites of the O−H···O protons are broad and almost equal, as can be seen in Figure 1b, implicating that any underlying interbond protonic hopping motion is slow on the millisecond time scale. This is also indicated by the fact that spin−lattice relaxation times of these two broad components were around 160 s for recrystallized powder without drying and around 230 s under vacuum pumping for 5 min and showed no temperature dependence within the experimental temperature range from 280 to 330 K. The derived inference is that the interbond proton hopping is very slow, in sharp contrast to an earlier proposed mechanism for LDP’s 1H superionic conductivity.11,12 Conductivity measurements were then undertaken to support the above conclusion from NMR that proton conductivity in LDP is strongly associated with the adsorbed H2O. Figure 3 shows the plots of the relative intensity (solid

Figure 2. Two-dimensional cross-polarization-based heteronuclear 1 H−31P correlation spectrum of recrystallized LDP powder.

directly coupled with phosphorus, as seen in the CP-based heteronuclear 31P−1H correlation spectrum while this narrow peak is not, indicating the phosphorus in the tetrahedral PO4 groups strongly couples with the protons in the two different H-bonding environments but does not have effective coupling with the highly dynamic water protons. Interestingly, the position and intensity of the sharp peak vary in proportion to the sample’s H2O content (controlled by the vacuum pumping time), showing that the H2O protons experience a fast exchange with the short O−H···O protons. Since the H2 of the short hydrogen bond appears at 12.4 ppm and the water peak normally at 4.7 ppm, the exchange rate has to be much larger than their chemical shift difference of 7.7 ppm, i.e., ≫4620 Hz at 600 MHz. In the fast exchange limit, the observed chemical shift δobs appears in the position given by the mean of the chemical shifts δA and δB from the two species participating in the fast exchange process, weighted by the their equilibrium concentrations pA and pB (pA + pB = 1):29 δobs = pA δA + pB δ B

Figure 3. Plots of 1H signal intensity (solid squares, left axis) of the narrow peak from the recrystallized powder and the DC electrical conductivity (solid circles, right axis) of the recrystallized powder pellet as a function of vacuum-pumping time. The 1H signal intensities were normalized to that from the unpumped sample.

squares, left axis) of the fast exchanged proton resonance of the recrystallized powder and the protonic conductivity (solid circles, right axis) as a function of vacuum pumping. It is seen that the intensity of the fast exchanging proton signal decreases with increasing pumping time, showing that this peak originates from the adsorbed water. Importantly, as can be seen in Figure 3, the DC proton conductivity of the LDP pellet also decreases with water depletion: 10−4 Ω−1 cm−1 for samples prepared in the ambient atmosphere versus 10−7 Ω−1 cm−1 for heavily pumped-upon samples. Similarly, the AC conductivity measurements clearly indicate a close correlation between the water content in the LDP pellet and the proton conductivity, as shown in Figure 4. Therefore, the adsorbed H2O in LDP plays a critical role in protonic conductivity. It is worth noting that the AC electrical conductivity of a single crystal measured at 1 Hz is about 1 × 10−7 Ω−1 cm−1 in the ambient atmosphere, while decreases to about 1 × 10−12 Ω−1 cm−1 after vacuum pumping for 1 h at RT, again indicating that the adsorbed water on the surface playing a critical role in

(1)

It is quite possible that the water protons exchange with both the long and the short hydrogen bonds, but the NMR spectra under high water content shift the water peak toward that from the short ones, well past that from the longer ones. Clearly the exchange is dominated by the protons with the short bond lengths; otherwise the water peak could never go past that from the ones with the long bond lengths. Unfortunately, the peak from the long bond length H’s is covered by the strong water 13389

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393

The Journal of Physical Chemistry C

Article

and not just hydrogen-bonded to each other. The condition of high protonic conductivity is also that of strong exchange between the water protons and the LDP hydrogens. Conductivity measurement on highly dehydrated samples provides the background conduction due to just proton hopping within the LDP’s O−H···O bonds, which is orders of magnitude too low compared to that shown by hydrated samples. Thus, the mechanism of proton conductivity in LDP must involve hopping between water protons and O−H···O protons, with the water acting as a catalyst for the intermolecular H hopping and conduction under applied electric fields. An important question is whether the observed protonic conductivity is due to just conduction via a surface layer of water or does the water act as a catalyst promoting the conductivity involving the hopping of the LDP hydrogens. To answer it, we note that LDP shows high conductivity of 10−4 Ω−1 cm−1 at RT and relative humidity of 50%. At this ambient condition, the moisture adsorbed by the pellet was ∼1 wt % or in terms of volume percentage, ∼2.5 vol %. For a given LDP pellet (1 cm2 area by 1 mm thickness), the resistance of the LDP pellet containing ∼1 wt % of moisture was measured to be ∼1 kΩ. To distinguish the contributions to the proton conductivity due to the interconnected water, LDP, and the interactions between water and LDP, we have measured and estimated the resistance due to each of these factors. We confirm that the resistance of just the interconnected water and just the LDP are very high compared to that of the LDP containing ∼1 wt % moisture. First, the resistance Rw due to the interconnected water in the pellet can be estimated by

Figure 4. Correlation of pellet mass (solid squares, left axis) (pressure for making the pellet: 4000 kg/cm2) with AC electrical conductivity (solid circles, right axis) at 10 Hz measured as a function of pumping time.

electrical conductivity of LDP. Figure 5 shows the measured 1H spin−lattice relaxation rates (1/T1) (for the adsorbed H2O)

Rw =

1 l σw A(2.5%)

(2)

where σw is the conductivity of distilled water which we have measured to be 7 × 10−6 Ω−1 cm−1 at RT, l is the thickness of the pellet, A is the area of the pellet, and 2.5% is the filling factor of the adsorbed moisture. From this, we calculated the resistance due to the interconnected water to be ∼560 kΩ. Second, we have measured the resistance of LDP at 0% relative humidity which is equal to the resistance of just the LDP crystals to be ∼1 × 1012 Ω at RT. Both resistances are very high compared to that of the LDP containing a minor amount of moisture. Therefore, the high proton conductivity of LDP is due to adsorbed water interacting with protons in LDP rather than purely interconnected surface water. 3.3. 31P and 7Li Measurements Support the New Mechanism of Protonic Conductivity. In order to gain further understanding of the conduction mechanism, we measured 31P and 7Li spectra and relaxation times. Figures 6 and 7 respectively show single-scan 31P and 7Li DP-MAS spectra at 21.2 T at a spinning speed of 13.5 kHz. Two peaks are seen in the 31P spectra; their relative intensity changes as the water is depleted via pumping on the sample. The smaller peak can be assigned to the minority phosphate groups with which the adsorbed water interacts, as the peak intensity decreases upon vacuum pumping. This smaller peak also exhibits a very short T1 value, less than 0.1 s. The main 31P peak has a very long (∼800 s) relaxation time. The water-dependent small peak, with a factor of 8000 shorter T1 than the tetrahedral PO4 groups in the LDP lattices, must originate from a very fast reorientation or rotation of the phosphate ions. Such fast reorientation/rotation is facilitated by the adsorbed water

Figure 5. 1H NMR spin−lattice relaxation rates (for the adsorbed H2O) fitted by using the Bloembergen−Purcell−Pound (BPP) model and the activation energies as a function of pumping time.

under different vacuum-pumping times as a function of temperature. The dashed lines were fitted using the Bloembergen−Purcell−Pound (BPP) model,30,31 and the derived activation energies as a function of pumping time are plotted in the inset of Figure 5. This fast lattice−water exchanged peak exhibits very short T1 values, as opposed to several hundred seconds for the protons in the hydrogen bonds. Interestingly, the lower activation energy of the T1 relaxation process in the larger H2O-containing LDP samples correlates with a higher protonic conductivity. Again, these data imply that fast exchange between the adsorbed H2O protons and those in the short O−H···O bonds is the primary mechanism underlying the superprotonic conductivity in LDP. The proton NMR chemical shift acts as a sensitive probe of the local structure of the adsorbed water. Used as a reference, bulk water would show a peak at 4.7 ppm. If the adsorbed water were in a sort of bulk-water form, the NMR peak from the adsorbed water would stay at 4.7 ppm regardless of the water content. The fact that the peak position moves as the water content is changed is a strong evidence that the water protons are exchanging rapidly with the H’s of the LDP hydrogen bonds 13390

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393

The Journal of Physical Chemistry C

Article

Figure 8. Schematic of the proton exchange model of LDP with the adsorbed H2O molecules.

bond in the short hydrogen bonds (H2) is 2.561 Å, the hydrogen bond between H2 and its oxygen is relatively weak, so that the H2 proton can be easily attracted by the nearby water molecules adsorbed by the H2PO4− anions and LiO4 tetrahedra. With the addition of the proton, the water molecule becomes H3O+, an ion contradicting the standard two-proton ice rule, thus resulting in pushing a proton away to the neighboring H2O molecule. As the H2 proton is attracted away by the water molecule, its place can be easily replaced by a proton from other water molecules, thus resulting in a fast exchange between the H-bonding proton H2 and the adsorbed water molecules. We conclude that this H2O proton exchange in concert with the rotational fluctuations of the PO4 moieties is the dominant mechanism of LDP’s exceptionally high protonic conductivity. The above results suggested that LDP’s conductivity could be modulated by controlling the amount of surface adsorbed water. This was indeed found to be the case. Figure 9 shows the

Figure 6. 31P DP-MAS NMR spectra of the recrystallized powder at 900 MHz with (bottom) and without (top) vacuum pumping. The deconvoluted peaks are shown in blue and green.

Figure 7. 7Li DP-MAS NMR spectra of the recrystallized powder at 900 MHz with (bottom) and without (top) vacuum pumping. The deconvoluted peaks are shown in blue and green.

molecules. In the CP-based two-dimensional heteronuclear 1 H−31P correlation spectrum of Figure 2, two broad 1H resonances from the long and short hydrogen bonding in the LDP lattice are shown to be directly coupled with phosphorus, while a small 31P peak (∼0 ppm) associated with adsorbed water is not shown because it does not have effective coupling with the two protons in the two different hydrogen bonds. Figure 7 shows single-scan 7Li spectra of recrystallized LDP powder under different conditions. Together with the main peak, we see a narrow component whose intensity decreases with increasing of vacuum-pumping time. This small and narrow component has a very short T1 (∼0.1 s), as compared to the main peak (∼1200 s). Again, it appears to arise from interaction between LiO4 and surrounding water molecules. The narrow and the broad peaks could be easily separated via deconvolution. It should be noted that the intensities of both the 31P and the 7Li peaks related to the adsorbed water are rather small as compared to the signals from the rest of the LDP lattice, indicating that only the tetrahedral PO4 groups and Li moieties in the surface area are interacting with the adsorbed water molecules. LiO4 tetrahedra are linked by their vertices forming [100] isolated chains23 and have weak interaction with surrounding water molecules. Therefore, LiO4 moieties are likely to have no direct contribution to LDP’s high protonic conductivity. Figure 8 shows the schematic of the proton exchange model of LDP with adsorbed water molecules. Since the inter-oxygen

Figure 9. Humidity-dependent proton conductivity with different pressures used for making pellets. Top figure is relative humidity (R.H.) vs time. The curves (red: 2000 kg/cm2; green: 4000 kg/cm2; blue: 6000 kg/cm2) show proton conductivity as a function of pressure.

conductivity of three different pressed LDP pellets (red: 2000 kg/cm2; green: 4000 kg/cm2; blue: 6000 kg/cm2) under cyclic humidity conditions at ambient pressures. For all samples, as the humidity is increased from 10% to 50%, the resulting conductivity increases by an order of magnitude. More importantly, when the humidity is cycled, the conductivity oscillates accordingly, implying that the physically adsorbed water is the driving force for the conductivity. The slow humidity response is consistent with the hydrogen bondingdriven water adsorption/desorption process. The conductivity 13391

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393

The Journal of Physical Chemistry C

■

can also be controlled by squeezing the water out of the powder during the pellet fabrication. For samples made under 2000, 4000, and 6000 kg/cm2 pressures, the water content was driven away by 7.8%, 10.9%, and 12.0%, respectively. As a consequence, the conductivity is lower for samples prepared under more strongly squeezed (more strongly pressed) samples, and vice versa. The proton conductivity of LDP can be well controlled by the adsorbed water content through humidity and pressure. We stress that the present work is distinctly different from the earlier studies on CDP13,14,22 and RDP,16 where humidity was used to control the sample dehydration at high temperatures close to 250 °C. The current study shows that LDP’s conductivity can be modulated reversibly which should add a new dimension to its potential applications in the fields of fuel cells and humidity sensors based on hydrogen-bonded proton conductors.

REFERENCES

(1) Norby, T. The Promise of Protonics. Nature 2001, 410, 877− 878. (2) Haile, S. M.; Chisholm, C. R. I.; Sasaki, K.; Boysen, D. A.; Uda, T. Solid Acid Proton Conductors: From Laboratory Curiosities to Fuel Cell Electrolytes. Faraday Discuss. 2007, 134, 17−39. (3) Kreuer, K.-D. Proton Conductivity: Materials and Applications. Chem. Mater. 1996, 8, 610−641. (4) Catti, M.; Ivaldi, G. Crystal Structure of LiH2PO4, Structural Topology and Hydrogen Bonding in the Alkaline Dihydrogen Orthophosphates. Z. Kristallogr. 1977, 146, 215−226. (5) Kweon, J. J.; Lee, K. W.; Lee, K.-S.; Oh, I.-H.; Lee, C. E. Charge Dynamics in KH2PO4 Systematically Modified by Proton Irradiation. J. Appl. Phys. 2011, 110, 044101. (6) Boysen, D. A.; Haile, S. M.; Liu, H.; Secco, R. A. HighTemperature Behavior of CsH2PO4 under Both Ambient and High Pressure Conditions. Chem. Mater. 2003, 15, 727−736. (7) Lee, K.-S.; Moon, J.; Lee, J.; Jeon, M. High-Temperature Phase Transformations in LiH2PO4 and Possible Solid-State Polymerization. Solid State Commun. 2008, 147, 74−77. (8) Kweon, J. J.; Lee, K. W.; Lee, C. E.; Lee, K.-S.; Jo, Y. J. Impedance Spectroscopy of the Superprotonic Conduction in LiH2PO4. Appl. Phys. Lett. 2012, 101, 012905. (9) Šušić, M. V.; Minić, D. M. Electric and Electrochemical Properties of Solid LiH2PO4. Solid State Ionics 1981, 2, 309−314. (10) Oh, I. H.; Lee, K.-S.; Meven, M.; Heger, G.; Lee, C. E. Crystal Structure of LiH2PO4 Studied by Single-Crystal Neutron Diffraction. J. Phys. Soc. Jpn. 2010, 79, 074606. (11) Kweon, J. J.; Lee, K. W.; Lee, C. E.; Lee, K.-S. Nuclear Magnetic Resonance Study of the Superprotonic Conduction in LiH2PO4. Appl. Phys. Lett. 2011, 98, 262903. (12) Kweon, J. J.; Lee, K. W.; Lee, K.-S; Lee, C. E. Rotating-Frame Nuclear Magnetic Resonance Study of the Superprotonic Conduction in LiH2PO4. Solid State Commun. 2013, 171, 5−7. (13) Kim, G.; Blanc, F.; Hu, Y.-Y.; Grey, C. P. Understanding the Conduction Mechanism of the Protonic Conductor CsH2PO4 by Solid-State NMR Spectroscopy. J. Phys. Chem. C 2013, 117, 6504− 6515. (14) Otomo, J.; Minagawa, N.; Wen, C.-J.; Eguchi, K.; Takahashi, H. Protonic Conduction of CsH2PO4 and Its Composite with Silica in Dry and Humid Atmospheres. Solid State Ionics 2003, 156, 357−369. (15) Yoshimi, S.; Matsui, T.; Kikuchi, R.; Eguchi, K. Temperature and Humidity Dependence of the Electrode Polarization in Intermediate-Temperature Fuel Cells Employing CsH2PO4/SiP2O7Based Composite Electrolytes. J. Power Sources 2008, 179, 497−503. (16) Li, Z.; Tang, T. High-Temperature Dehydration Behavior and Protonic Conductivity of RbH2PO4 in Humid Atmosphere. Mater. Res. Bull. 2010, 45, 1909−1915. (17) Chandra, S.; Kumar, A. Investigations on the Proton Transport Mechanism in Potassium Dihydrogenphosphate. J. Phys.: Condens. Matter 1991, 3, 5271−5286. (18) Pinto, N. J.; Kahol, P. K.; McCormick, B. J.; Dalal, N. S.; Wan, H. Charge Transport and Electron Localization in Polyaniline Derivatives. Phys. Rev. B 1994, 49, 13983−13986. (19) Zuo, F.; Angelopoulos, M.; MacDiarmid, A. G.; Epstein, A. J. Transport Studies of Protonated Emeraldine Polymer: A Granular Polymeric Metal System. Phys. Rev. B 1987, 36, 3475−3478. (20) Nogami, M.; Abe, Y. Evidence of Water-Cooperative Proton Conduction in Silica Glasses. Phys. Rev. B 1997, 55, 12108−12112. (21) Nogami, M.; Nagao, R.; Cong, W.; Abe, Y. Role of Water on Fast Proton Conduction in Sol-Gel Glasses. J. Sol-Gel Sci. Technol. 1998, 13, 933−936. (22) Boysen, D. A.; Uda, T.; Chisholm, C. R. I.; Haile, S. M. HighPerformance Solid Acid Fuel Cells Through Humidity Stabilization. Science 2004, 303, 68−70. (23) Lee, K.-S.; Oh, I.-H.; Kweon, J. J.; Lee, C. E.; Ahn, S.-H. Crystal Growth and Morphology of LiH2PO4. Mater. Chem. Phys. 2012, 136, 802−808.

4. CONCLUSIONS Correlated high-field, high-resolution solid-state NMR and electrical conductivity measurements on the superionic conductor LDP show that the mechanism of its protonic conductivity is the exchange of H2O protons mainly with its short hydrogen bonds and not the interbond hopping of protons between the O−H···O bonds. While water is known to play a role in other hygroscopic conducting systems, such as some polymers like polyanilines,18 the current work is the first report on the role of water in proton conduction in the LDP type hydrogen-bonded solids, in both single crystal and powder form. In fact, the presently proposed mechanism is a significant improvement over an earlier proposed model11,12 wherein the role of water was not recognized. Finally, we have shown that LDP’s protonic conductivity can be modulated reversibly by factors of several orders of magnitude via controlled exposure to water, a result of high relevance in the utility of LDP in hydrogen fuel cells and related applications, and possibly for proton conduction in other classes of hydrogen-bonded solids.

■

Article

AUTHOR INFORMATION

Corresponding Authors

*Tel 850-644-3398; Fax 850-644-8281; e-mail [email protected]. edu (N.S.D.). *Tel 850-644-5044; Fax 850-644-1366; e-mail [email protected]. edu (R.F.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We are grateful for financial support from the User Collaboration Grants Program (UCGP) 5084 at the National High Magnetic Field Laboratory (NHMFL) which is supported by National Science Foundation (NSF) Cooperative Agreement No. DMR-1157490, the State of Florida, and the U.S. Department of Energy. E.S. is supported by NSF-DMR 1005293 and 1309146. We thank Prof. Jim P. Zheng at the College of Engineering at FSU and Prof. James Brooks at NHMFL for permitting us to use their conductivity equipment and Jason Kitchen and William W. Brey at NHMFL for helpful discussion on variable temperature setups. 13392

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393

The Journal of Physical Chemistry C

Article

(24) Lee, K.-S.; Ko, J.-H.; Moon, J.; Lee, S.; Jeon, M. Raman Spectroscopic Study of LiH2PO4. Solid State Commun. 2008, 145, 487−492. (25) Voronov, A. P.; Babenko, G. N.; Puzikov, V. M.; Iurchenko, A. N. Growth of LiH2PO4 Single Crystals from Phosphate Solutions. J. Cryst. Growth 2013, 374, 49−52. (26) Steven, E.; Park, J. G.; Paravastu, A.; Lopes, E. B.; Brooks, J. S.; Englander, O.; Siegrist, T.; Kaner, P.; Alamo, R. G. Physical Characterization of Functionalized Spider Silk: Electronic and Sensing Properties. Sci. Technol. Adv. Mater. 2011, 12, 055002. (27) Eckert, H.; Yesinowski, J. P.; Silver, L. A.; Stolper, E. M. Water in Silicate Glasses: Quantitation and Structural Studies by Proton Solid Echo and Magic Angle Spinning NMR Methods. J. Phys. Chem. 1988, 92, 2055−2064. (28) Oh, I.-H.; Kweon, J. J.; Oh, B. H.; Lee, C. E. Proton Conduction in KH2PO4 Modified by Proton Irradiation. J. Korean Phys. Soc. 2008, 53, 3497−3499. (29) Levitt, M. H. Spin Dynamics, Basics of Nuclear Magnetic Resonance; John Wiley & Sons, Ltd.: London, 2001. (30) Slichter, C. P. Principles of Magnetic Resonance; Springer: Berlin, 1990. (31) Abragam, A. Principles of Nuclear Magnetism; Oxford University Press: Oxford, 1983.

13393

dx.doi.org/10.1021/jp501531h | J. Phys. Chem. C 2014, 118, 13387−13393