Fast Li-Ion Dynamics in Stoichiometric Li2S–Ga2Se3–GeSe2 Glasses

Sep 25, 2017 - The same data were used to generate the conductivity vs frequency (Bode) plots at various temperatures for each sample. 7Li NMR Spectro...
0 downloads 10 Views 1MB Size
Article Cite This: Chem. Mater. 2017, 29, 8704-8710

pubs.acs.org/cm

Fast Li-Ion Dynamics in Stoichiometric Li2S−Ga2Se3−GeSe2 Glasses Maxwell A.T. Marple,† Bruce G. Aitken,‡ Sangtae Kim,† and Sabyasachi Sen*,† †

Department of Materials Science and Engineering, University of California at Davis, Davis, California 95616, United States Science & Technology Division, Corning Inc., Corning, New York 14831, United States



ABSTRACT: The temperature dependence of Li-ion transport is explored in novel stoichiometric chalcogenide glasses in the ternary system Li2S−Ga2Se3− GeSe2 using electrochemical impedance spectroscopy (EIS) and 7Li nuclear magnetic resonance (NMR) spectroscopy. The dc conductivity in these glasses monotonically increases with Li2S content with the highest room-temperature conductivity being ∼10−4 S/cm in the glass of composition: 50% Li2S−10% Ga2Se3−40% GeSe2 with an activation energy of 0.37 eV. Analysis of the 7Li NMR central transition line shape reveals that Li ions are randomly distributed in the structure of these glasses and their rapid hopping transport is responsible for the electrical conductivity. The excellent agreement between the experimentally measured dc conductivity and that calculated from the Li-ion hopping frequency obtained from EIS and NMR implies that the ionic transference number in all glasses is close to unity and these glasses behave as strong electrolytes.



requirement of Ge and Ga.18 The Li2S content in the LGGS glasses with R > 1 is sufficiently high to create percolation channels of NBC where Li-ion mobility is expected to be high. Recent studies have indicated that the “mixed anion” sulfoselenide glasses containing both S and Se may have superior ionic conductivity, although attempts in synthesizing Li-ion conducting sulfoselenide glassy electrolytes have resulted in glasses that are unstable against extensive crystallization.19,20 Here we report the results of a comprehensive study of the ionic conductivity and Li-ion mobility in sulfoselenide glasses in the LGGS system using a combination of electrochemical impedance spectroscopy (EIS) and variable temperature 7Li nuclear magnetic resonance (NMR) spectroscopy. A comparison between the correlation times obtained from EIS and NMR line width analysis is used to establish a direct mechanistic link between the length scale and time scale of the Li-ion hopping dynamics and the macroscopic electrical conductivity in these complex glasses. We report the existence of glass compositions in the LGGS system that meets the benchmark of minimum room-temperature ionic conductivity of ∼10−4 S/cm, which could be potentially optimized for consideration for future device use.

INTRODUCTION Fast ion-conducting solid electrolytes are the basis of a wide range of emergent technologies in the areas of energy storage, display systems, and sensing.1−6 There is particular interest in utilizing solid electrolytes in next generation solid state lithium batteries which have the benefit of being less prone to combustion upon impact or exposure to ambient atmosphere and offer new chemistries for cathode and anode pairs to increase the electrical power and energy density of the cell such as using Li/S electrodes.7,8 Furthermore, glassy solid electrolytes exhibit several promising properties such as high ionic transference number and increased stability to Li metal by hindering growth of Li dendrites.9,10 Glassy solid electrolytes are also uniquely advantageous for their compositional flexibility and for their potential to stabilize “superionic” crystalline phases encapsulated in inert glassy matrix, forming a glass−ceramic.8,11−15 Compared to glassy oxides, chalcogenide glassy electrolytes are particularly promising for their high ionic conductivity and superior ductility allowing for room temperature processing.9,16,17 However, the search continues for novel chalcogenide solid electrolytes with optimum combination of high room-temperature conductivity (10−4 S/ cm) and improved chemical stability. In a recent work we reported glass formation in the stoichiometric system Li2S−Ga2Se3−GeSe2 (LGGS).18 The structure of these novel LGGS glasses is characterized by a network comprised of corner-sharing (Ge,Ga)SxSe4−x tetrahedral units. In these glasses Li serves as a charge compensator for the [GaSxSe4−x]−1 tetrahedra.18 Li in glasses with R > 1 (R = Li2S/Ga2Se3) acts as a network modifier, forming nonbridging chalcogens (NBC) that disrupt the connectivity of the network. On the other hand, glasses with R < 1 contain homopolar Ge− Ge linkages that form as a consequence of these glasses being chalcogen deficient, in order to satisfy the 4-fold coordination © 2017 American Chemical Society



METHODS

Synthesis. The synthesis of 7 LGGS glasses (see Table 1 for compositions) was carried out using a two-step quenching method within evacuated fused quartz ampules.18 Approximately 25 g batches of the precursor Ga−Ge−Se alloys were prepared for each composition by melting mixtures containing appropriate amounts of the constituent elements Se, Ga, and Ge (99.9999% purity) in a Received: July 9, 2017 Revised: September 24, 2017 Published: September 25, 2017 8704

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710

Article

Chemistry of Materials Table 1. Density, Room-Temperature dc Conductivity σRT, and Corresponding Activation Energy Edc for all LGGS Glasses in This Study Li2S−Ga2Se3−GeSe2 (mol %) 40−00−60 40−10−50 45−10−45 50−10−40 7.5−32.5−60 15−25−60 20−30−50

density (g/cm3) 3.65 3.75 3.69 3.56 4.376 4.253 4.279

σRT (S/cm) 6.6 8.9 3.8 7.5 1.5 1.5 2.1

× × × × × × ×

−6

10 10−6 10−5 10−5 10−7 10−6 10−6

Avance 500 (11.7 T) spectrometer operating at a Larmor frequency of 194.4 MHz for 7Li. The samples were packed inside a glovebox into ZrO2 rotors equipped with Macor caps with Viton o-rings to prevent exposure to air during data collection. A solid (quadrupole) echo pulse sequence of π/2−tp−π/2 with the interpulse delay tp = 60 μs and a recycle delay of 10 to 25 s was used to collect each static 7Li spectrum. The π/2 pulse was calibrated to 2 μs. The sample temperature was controlled by flowing dry N2 gas through a heat exchanger submerged in liquid N2 and heated with a 4 mm triple resonance probe. The temperature was calibrated to within ±2 K using the known temperature dependence of the chemical shift difference between the 1H NMR signals from the CH2 and the OH sites in ethylene glycol.23 The 7Li chemical shifts were externally referenced to the isotropic chemical shift δiso of solid LiCl (δiso = 0 ppm).

Edc (eV) 0.34 0.36 0.36 0.37 0.53 0.47 0.45



rocking furnace for 3 days at 950 °C and subsequently quenched in water. This precursor material was extracted and mixed with appropriate amounts of Li2S (99.99% purity) to make approximately 10 g batches for each composition. These batches were melted in silicon coated fused quartz ampules. The ampules were sealed under vacuum at 8 × 10−5 Torr and loaded into a vertical furnace at 950 °C. After 15 min of melting, the ampules were removed and tilted to homogenize the liquid and then loaded back into the furnace to melt for another 15 min. The ampules were subsequently quenched in water and annealed in a furnace at 200 °C for 30 min. All R > 1 glasses are quite hygroscopic and react immediately upon exposure to moisture to release H2S/H2Se, thus requiring general handling within a moisture free Ar/N2-filled glovebox. The R < 1 glasses are more stable, showing only some reaction with moisture after 24 h of continuous exposure. Density. The densities of the R > 1 glasses were measured using the Archimedes’ method with toluene as the immersion medium. All measurements were carried out inside an Ar-filled glovebox with moisture levels below 0.8 ppm and at ambient temperature. Appropriate corrections were made for the Ar atmosphere and the temperature and pressure in calculating the densities.21,22 The reported density for each of the R > 1 glasses is an average of 6 consecutive measurements on about 1 g of the glass sample and is determined to within ±0.02 g/cm3. Densities of the R < 1 glasses were measured using a Micromeritics AccuPyc II gas expansion pycnometer under a helium environment of 6 N purity. Approximately 1 to 2 g of each sample was loaded into a 1 cm3 cup. Reported densities are averages of 10 consecutive measurements at 20 °C and are determined to within ±0.003 g/cm3. Electrochemical Impedance Spectroscopy. Bulk glass samples were polished down to rectangular shapes, and both faces were coated in an isopropanol based graphite paste for the electrodes and allowed to cure. The samples were loaded into an airtight brass cell composed of two brass plates sandwiching a Teflon body and silicone gaskets. A spring and floating brass electrode was used to ensure good contact between the brass and the carbon electrodes. This containment cell was placed within a Nalgene jar modified to have airtight gas flow and electrical feed-through. The assembly was immersed into a 50/50 ethylene glycol/water mixture in a temperature-controlled circulator bath. During data collection, temperature regulated Ar gas was flowed through the cell, and data points were collected by alternating heating and cooling through the temperature range of measurement −25 °C to 60 °C. A Novocontrol impedance analyzer was used to measure the impedance through a frequency range of 1 Hz−10 MHz. The Cole− Cole plots showed a single depressed semicircle and a low frequency tail due to electrode polarization. The dc conductivity was determined from these plots by simulating them with a single parallel R−Q circuit where R is a resistor element and Q is a constant phase element. The same data were used to generate the conductivity vs frequency (Bode) plots at various temperatures for each sample. 7 Li NMR Spectroscopy. The temperature dependence of the 7Li NMR line shape was investigated in three LGGS glasses (40% Li2S− 60% GeSe2, 15% Li2S−25% Ga2Se3−60% GeSe2, 50% Li2S−10% Ga2Se3−40% GeSe2) that were chosen to highlight the influence of Li2S content as well as the effect of incorporation of Ga2Se3 on the Liion dynamics. All 7Li NMR experiments were carried out on a Bruker

RESULTS Density. The measured densities of the LGGS glasses are summarized in Table 1. The corresponding variation in the molar volumes of these glasses are shown in Figure 1 as a

Figure 1. Variation of molar volume of LGGS glasses with Li2S content. Inset shows Li−Li separation distance obtained from the molar volume data under the assumption of a random distribution of Li. See text for details.

function of the Li2S content. The molar volume decreases monotonically with increasing Li2S content regardless of other compositional parameters such as R or Ga2Se3/GeSe2 ratio. The average Li−Li separation distance in the structure of these glasses can then be calculated from the molar volume under the assumption of a random spatial distribution of Li in the glass structure. The results of this calculation are shown in Figure 1. Electrochemical Impedance Spectroscopy. The temperature dependence of the dc conductivity of all LGGS glasses as obtained from the EIS data is shown in Figure 2. The dc conductivity follows an Arrhenius behavior for all compositions over the temperature range of the present measurements (−25 to 60 °C). The corresponding activation energies Edc are summarized in Table 1. Replacement of GeSe2 with Ga2Se3 results in only a minor difference in conductivity between the binary 40% Li2S−60% GeSe2 and ternary 40% Li2S−10% Ga2Se3−50% GeSe2 glasses. Although the dc conductivity of these glasses monotonically increases with increasing Li content, the activation energy initially drops relatively rapidly for the R < 1 glasses while further increase in Li content beyond 40% Li2S results in little change for the R > 1 glasses (Figure 3). As the Li2S content increases by about a factor of 7, there is nearly 3 orders of magnitude increase in the room temperature conductivity (Figure 3), which implies considerable change in 8705

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710

Article

Chemistry of Materials

Figure 2. Arrhenius plot of dc conductivity for all LGGS glasses in this study. Glass compositions are given alongside each curve, in the form of three numbers separated by dashes which correspond from left to right to mol % Li2S, Ga2Se3, and GeSe2, respectively. Dashed straight lines through the data points are linear least-sqaures fits with the vertical dashed black line indicating room temperature ∼ 25 °C.

the Li-ion mobility in these glasses with increasing modification of the network. The highest room-temperature dc conductivity is found in the 50 mol % Li2S glass composition, reaching a value of ∼7.5 × 10−5 S/cm with an activation energy of ∼0.37 eV, placing it on the edge of the typical considerations for classification as a “superionic” conductor.24 Additional information about the dynamical process responsible for the electrical conductivity can be obtained from the frequency dependence of conductivity, which is shown in Figure 4 in the form of Bode plots for three representative compositions at room temperature. The frequency independent conductivity plateau in these Bode plots marks the “dc conductivity” σdc of the bulk glass. At high frequencies, the dispersive nature of conductivity becomes apparent, i.e., the conductivity σ(ν) becomes strongly dependent on the frequency of measurement ν. This frequency dependence of the conductivity σ is typically described in the literature by the α⎤ ⎡ ν Jonscher expression: (ν) = σdc⎢1 + ν ⎥, where 0 ≤ α ≤ 1 ⎣ ⎦ H is a dimensionless constant and νH is the frequency marking the onset of frequency-dependent conductivity where σ(ν) = 2·σdc.25−28 This crossover frequency νH has been identified in the random barrier model (RBM) of ionic conduction in glasses with the inverse of the time required for the mobile ions to overcome the percolation barrier in the energy landscape.29 It is clear from Figure 4 that νH generally increases with Li2S content in LGGS glasses suggesting a corresponding increase in the mobility of the Li ions in the same direction. 7 Li NMR Line Shape Analysis. The temperature dependent evolution of the static 7Li NMR solid-echo spectra for the 40% Li2S−60% GeSe2 glass is shown in Figure 5. In all cases the 7 Li NMR line shape at the lowest temperature is characterized by a narrow central Gaussian line corresponding to the central transition superimposed on a broad, low-intensity component corresponding to the satellite transitions. It is well-accepted in

Figure 3. Room-temperature dc conductivity (a) and activation energy Edc (b) of ternary LGGS glasses with R < 1 (triangles) and R > 1 (circles). Corresponding data for the binary 40% Li2S−60% GeSe2 glass are shown with black squares.

the literature that the primary source of broadening of the 7Li central transition line shape in Li-rich crystals and glasses is from dipolar interactions, while the broadening for the satellite transitions is largely controlled by the quadrupolar interaction.30 Here we focus on the 7Li central transition line shape as its progressive narrowing with increasing temperature can be followed with high precision and can be directly linked to the long-range diffusion of Li ions that leads to the dynamical averaging of the dipolar broadening interaction. It is clear from Figure 5 that the width of the central transition line in the 7Li NMR spectra decreases monotonically with increasing temperature, indicating motional narrowing. The corresponding variation in the full width at half-maximum (fwhm) of the central transition line with temperature is shown for all three glasses in Figure 6. The fwhm at the lowest temperatures increases nearly linearly with increasing Li concentration, which is consistent with the expectation that the primary controlling factor for the central transition line width in these glasses is the 7Li−7Li homonuclear dipolar interaction and implies a random distribution of the Li-ions.31 The large gyromagnetic ratio (10.4 × 10−7 rad/Ts) and high natural abundance (92.6%) of 7Li in combination with the low natural abundance of 77Se (7.6%) and 73Ge (7.8%) as well as the small gyromagnetic ratio of the latter (−0.94 × 10−7 rad/ Ts) ensure that only the homonuclear 7Li−7Li dipolar coupling is significant in this system. Although the contribution from 7 Li−69,71Ga heteronuclear dipolar coupling to the fwhm cannot

( )

8706

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710

Article

Chemistry of Materials

Figure 4. Frequency dependence of conductivity for three representative LGGS glasses at room temperature. Glass compositions are denoted as in Figure 2. Dashed vertical lines indicate the location of the ac−dc crossover frequency νH. Inset shows variation of νH at room temperature with Li2S content for all LGGS compositions. Meaning of the symbols in the inset is the same as in Figure 3.

Figure 5. Right: Temperature dependence of the static 7Li NMR spectra of 40% Li2S−60% GeSe2 glass. CT and ST denote central transition and satellite transition, respectively. Left: Expanded view of the spectra to the right showing the central transition line shape.

Figure 6. Temperature dependence of the fwhm of the static 7Li NMR central transition line shape for three representative LGGS compositions. Glass compositions are denoted as in Figure 2.

broadening interaction (Figure 6). A complete averaging of the dipolar interaction in the motionally narrowed regime is also signified by the fact that the fwhm of the central transition line for all three compositions is nearly identical in this regime (Figure 6). An approximate measure of the correlation time τNMR for Li-ion hopping at the temperature corresponding to the inflection point of the nearly sigmoidal fwhm vs temperature curve in Figure 6 can be obtained from the relation τNMR (s) = (2πνRL)−1, where νRL is the rigid-lattice fwhm in Hz.30 Similarly, a rough estimate of the activation energy for Li-ion hopping ENMR can be obtained from the relation ENMR (eV) = (1.62 × 10−3)Tonset, where Tonset is the temperature of onset of line narrowing.30 These estimated values of τNMR and ENMR are listed in Table 2.

be neglected in Ga containing compositions, such contributions would be significantly smaller than that from the 7Li−7Li dipolar coupling. The initial increase in temperature does not affect this “rigid lattice” line width significantly as the hopping frequency of the Li ions remains small compared to the strength of the 7Li−7Li dipolar interaction. Further increase in temperature results in a relatively rapid lowering of the fwhm as the hopping frequency increases and becomes comparable to the dipolar interaction, resulting in a dynamical averaging of the latter (Figure 6). Finally, near room-temperature and above, the fwhm reaches a plateau in the “motionally narrowed” regime as the Li-ion hopping becomes fast enough to completely average the dipolar 8707

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710

Article

Chemistry of Materials Table 2. Comparison of the Crossover Frequency νH and the Inverse of the Correlation Time τNMR, Obtained from 7Li NMR Line Shape Analysis, at Specific Temperatures Indicated within Parenthesesa Li2S−Ga2Se3−GeSe2 (mol %)

νH (T) (±2 kHz)

τNMR−1 (T) (kHz)

Edc (eV)

ENMR (±0.02 eV)

50−10−40 40−0−60 15−25−60

12 (230 K) 17 (250 K) 18 (285 K)

29 (230 K) 26 (250 K) 15 (285 K)

0.37 0.34 0.47

0.30 0.32 0.38

jumps of Li ions. Therefore, the successful hops could be compound in nature; i.e., they may consist of multiple elementary hops if the average Li−Li intersite distances in a glass with low Li content is significantly longer than the length scale of an elementary hop, which is typically ∼2−3 Å. As the Li concentration increases, the Li−Li distance will eventually decrease to the distances characteristic of the elementary hop and a static percolation path will be established. In this scenario, the dc conductivity σdc can be calculated using the using the Nernst−Einstein (N−E) relation for a random walk model of vacancy diffusion that predicts

a

The dc conductivity activation energy Edc is also compared with the corresponding activation energy obtained from 7Li NMR line shape. See text for details.



σdc =

DISCUSSION The rapid rise in conductivity by nearly 3 orders of magnitude upon increasing the Li2S concentration by only a factor of 7 (Figure 2) clearly suggests that the Li-ion mobility must also increase rapidly with concentration in these LGGS glasses. This hypothesis is corroborated by the increase in the crossover frequency νH with increasing Li concentration and is consistent with the concomitant lowering of Edc in the R < 1 glasses (Figures 3 and 4). The relatively rapid lowering of Edc with increasing Li concentration (Figure 3) at low Li content has been observed in other glass-forming systems and can be related to a progressive shortening of the average Li−Li separation (Figure 4) and therefore the effective hopping distance.32−34 In contrast, the high Li content of the R > 1 glasses results in the formation of percolative conduction pathways through the glass network. The plateau of the Edc in Figure 3 is a manifestation of these percolated conduction pathways as they provide a channel of low energy barriers for Li motion that is mostly compositionally invariant once such a pathway is established.35 Further insight can be gained from a quick comparison between the crossover frequency νH obtained from the Bode plot and the hopping frequency τNMR−1 obtained from the 7Li NMR line width analysis as well as that between the corresponding activation energies Edc and ENMR (Table 2). It may be noted here that the estimation method used in this study to obtain ENMR is expected to underestimate its value.30 In spite of the approximate nature of the estimates of τNMR and ENMR, the agreement between the NMR and EIS derived frequencies and activation energies, within their respective errors, is notable (Table 2). This result clearly implies that the hopping dynamics of Li ions must be solely responsible for the dc conductivity in these glasses; i.e., the ionic transference number must be close to unity. Previous molecular dynamics studies of ionic motion in glasses have shown that the mobile ions at short times remain mostly caged and perform forward−backward correlated hops. However, every so often a random fraction of these localized ions performs accelerated dynamics which take the ions to their next caging location and leads to long-distance diffusion and dc conductivity.36 Within the framework of the RBM, this longdistance diffusion only happens if an ion has sufficient energy to overcome the percolation barrier/threshold in an energy landscape with random barrier heights and νH corresponds to the frequency of such events that signify a “successful” hop.29 The close correspondence between νH and τNMR−1 (Table 2) then indicates that such successful hops associated with longdistance diffusion are also primarily responsible for an efficient averaging of the 7Li−7Li dipolar coupling. It is tempting to argue that these successful hops involve intersite (or intercage)

zi 2e 2nγa 2νH 6kT

with z, e, k, and T being charge number, elementary charge, Boltzmann constant, and absolute temperature, respectively; γ is a correlation factor, and a and νH are, respectively, the length scale and the frequency of a successful hop with a being equal to the Li−Li separation distance; and n is the number density of mobile ions.36 Ideally the N−E relation can be written as z 2e 2n σdc = i D*, where D* is the self (tracer) diffusion coefficient kT

of the mobile ion. However, in reality, ionic conductors often z 2e 2n

violate this relation such that σdc = ikT Dσ , which is captured by the correlation factor γ = D*/Dσ. The correlation factor 0 < γ ≤ 1 is also known as the Haven ratio in the literature, and its departure from unity is widely accepted to represent cooperative or correlated motion of ions.36 This departure of γ from unity results from the fact that, unlike the conductivity correlation function Cσ(t ) ∝ ∑ij ⟨vi(0)vj(t )⟩, the correlation function for self-diffusion CD(t) does not contain crosscorrelation terms i ≠ j since CD(t) = ⟨ri(0)ri(t)⟩.36 In these relations, ri and vi represent the position and the velocity of the mobile ions, respectively. Therefore, γ = 1 for completely uncorrelated motion while 0 < γ < 1 for correlated motion. Fast ion conductors with large concentration of mobile ions may be characterized by strongly correlated motion, and γ could be as low as ∼0.2 to 0.3 in these cases.36 Here, we use the N−E relation to calculate σdc at ambient temperature under the assumptions that (i) these glasses behave as strong electrolytes where all Li ions are mobile at time scales longer than νH−1 and (ii) in the absence of data on D*, we approximate γ = 1. It is noted that these approximations may result in an overestimation of σdc; however, the magnitude of this overestimation would be much smaller than the total range of σdc observed in these glasses (Figure 2). The σdc, thus calculated, is compared with the corresponding experimental results in Figure 7. Since νH by definition has the same activation energy as that of σdc, the excellent agreement observed in Figure 7 between the experimental and the calculated σdc at ambient temperature would imply a similar level of agreement at all temperatures. This result, therefore, provides strong support in favor of the physical interpretation of the crossover frequency νH presented above and is consistent with the assumption that these glasses behave as strong electrolytes. It should be noted here that a similar interpretation of a and νH within the N−E equation has been noted before34 and was experimentally unequivocally demonstrated in recent studies on ionic conduction in crystalline solid oxide electrolytes.37,38 8708

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710

Article

Chemistry of Materials

(3) Bachman, J. C.; Muy, S.; Grimaud, A.; Chang, H. H.; Pour, N.; Lux, S. F.; Paschos, O.; Maglia, F.; Lupart, S.; Lamp, P.; Giordano, L.; Shao-Horn, Y. Inorganic Solid-State Electrolytes for Lithium Batteries: Mechanisms and Properties Governing Ion Conduction. Chem. Rev. 2016, 116, 140−162. (4) Farahani, H.; Wagiran, R.; Hamidon, M. N. Humidity sensors principle, mechanism, and fabrication technologies: a comprehensive review. Sensors 2014, 14, 7881−7939. (5) Vassilev, V. S.; Boycheva, S. V. Chemical sensors with chalcogenide glassy membranes. Talanta 2005, 67, 20−27. (6) Brow, R. K.; Schmitt, M. L. A survey of energy and environmental applications of glass. J. Eur. Ceram. Soc. 2009, 29, 1193−1201. (7) Janek, J.; Zeier, W. G. A solid future for battery development. Nat. Energy 2016, 1, 16141. (8) Tatsumisago, M.; Hayashi, A. Development of Glass-Based Solid Electrolytes for Lithium-Ion Batteries. In Nanoscale Technology for Adanced Lithium Batteries; Osaka, T., Ogumi, Z., Eds.; Springer: New York, 2014; pp 63−80. (9) Agostini, M.; Aihara, Y.; Yamada, T.; Scrosati, B.; Hassoun, J. A lithium−sulfur battery using a solid, glass-type P2S5−Li2S electrolyte. Solid State Ionics 2013, 244, 48−51. (10) Ohtomo, T.; Hayashi, A.; Tatsumisago, M.; Tsuchida, Y.; Hama, S.; Kawamoto, K. All-solid-state lithium secondary batteries using the 75Li2S·25P2S5 glass and the 70Li2S·30P2S5 glass−ceramic as solid electrolytes. J. Power Sources 2013, 233, 231−235. (11) Tatsumisago, M.; Nagao, M.; Hayashi, A. Recent development of sulfide solid electrolytes and interfacial modification for all-solidstate rechargeable lithium batteries. J. Asian Ceram. Soc. 2013, 1, 17− 25. (12) Cao, C.; Li, Z.-B.; Wang, X.-L.; Zhao, X.-B.; Han, W.-Q. Recent Advances in Inorganic Solid Electrolytes for Lithium Batteries. Front. Energy Res. 2014, 2, 10.3389/fenrg.2014.00025 (13) Angulakshmi, N.; Stephan, A. M. Efficient Electrolytes for Lithium−Sulfur Batteries. Front. Energy Res. 2015, 3, 10.3389/ fenrg.2015.00017 (14) Seino, Y.; Ota, T.; Takada, K.; Hayashi, A.; Tatsumisago, M. A sulphide lithium super ion conductor is superior to liquid ion conductors for use in rechargeable batteries. Energy Environ. Sci. 2014, 7, 627−631. (15) Kim, S. K.; Mao, A.; Sen, S.; Kim, S. Fast Na-Ion Conduction in a Chalcogenide Glass−Ceramic in the Ternary System Na2Se− Ga2Se3−GeSe2. Chem. Mater. 2014, 26, 5695−5699. (16) Yamada, T.; Ito, S.; Omoda, R.; Watanabe, T.; Aihara, Y.; Agostini, M.; Ulissi, U.; Hassoun, J.; Scrosati, B. All Solid-State Lithium-Sulfur Battery using a Glass-Type P2S5-Li2S Electrolyte: Benefits on Anode Kinetics. J. Electrochem. Soc. 2015, 162, A646− A651. (17) Sakuda, A.; Hayashi, A.; Tatsumisago, M. Sulfide solid electrolyte with favorable mechanical property for all-solid-state lithium battery. Sci. Rep. 2013, 3, 2261. (18) Marple, M. A. T.; Aitken, B. G.; Sen, S. Synthesis and structural characterization of stoichiometric Li-Ga-Ge Sulfo-selenide glasses. J. Non-Cryst. Solids 2017, 457, 44−51. (19) Kim, J.; Yoon, Y.; Eom, M.; Shin, D. Characterization of amorphous and crystalline Li2S-P2S5-P2Se5 solid electrolytes for allsolid-state lithium ion batteries. Solid State Ionics 2012, 225, 626−620. (20) Liu, Z.; Tang, Y.; Wang, Y.; Huang, F. High performance Li2SP2S5 solid electrolyte induced by selenide. J. Power Sources 2014, 260, 264−267. (21) Gilgen, R.; Kleinrahm, R.; Wagner, W. Measurement and correlation of the (pressure, density, temperature) relation of argon I. The homogeneous gas and liquid regions in the temperature range from 90 to 340 K at pressures up to 12 MPa. J. Chem. Thermodyn. 1994, 26, 383−398. (22) Glen, N. F.; Johns, A. I. Determination of the density of toluene in the range from (293 to 373) K and from (0.1 to 30) MPa. J. Chem. Eng. Data 2009, 54, 2538−2545. (23) Ammann, C.; Meier, P.; Merbach, A. E. A Simple Multinuclear NMR Thermometer. J. Magn. Reson. 1982, 46, 319−321.

Figure 7. Comparison between experimental dc conductivity at room temperature from EIS and corresponding values calculated using the N−E relation, for all LGGS compositions. Dashed line along the diagonal corresponds to equality between experimental and calculated values. See text for details.



CONCLUSION The dynamics of Li ions in LGGS glasses are studied using a combination of EIS and 7Li NMR line shape analysis. The dc conductivity data indicate a rapid rise in Li-ion mobility with increasing temperature and, more interestingly, with Li concentration. The compositional variation of Edc indicates the formation of a low-energy barrier (∼0.35 eV) percolation pathway for Li-ion hopping through the glass network. The inverse of the correlation time scale for this hopping, as obtained from 7Li NMR line shape analysis, corresponds well with the ac−dc crossover frequency, obtained from EIS measurements. Equating the crossover frequency νH with the frequency of a successful hop over a length scale equal to the average Li−Li separation distance leads to robust estimates of σdc in all glasses, implying a strong-electrolyte type behavior at time scales longer than νH−1 with uncorrelated hopping dynamics of the Li ions.



AUTHOR INFORMATION

ORCID

Sangtae Kim: 0000-0001-6259-5132 Sabyasachi Sen: 0000-0002-4504-3632 Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by a GOALI grant from the National Science Foundation (NSF-DMR 1505185) to S.S. REFERENCES

(1) Patel, K. J.; Bhatt, G. G.; Ray, J. R.; Suryavanshi, P.; Panchal, C. J. All-inorganic solid-state electrochromic devices: a review. J. Solid State Electrochem. 2017, 21, 337−347. (2) Kim, J. G.; Son, B.; Mukherjee, S.; Schuppert, N.; Bates, A.; Kwon, O.; Choi, M. J.; Chung, H. Y.; Park, S. A review of lithium and non-lithium based solid state batteries. J. Power Sources 2015, 282, 299−322. 8709

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710

Article

Chemistry of Materials (24) Goodenough, J. B.; Kim, Y. Challenges for Rechargeable Li Batteries. Chem. Mater. 2010, 22, 587−603. (25) Almond, D. P.; West, A. R. Mobile ion concentration in solid electrolytes from an analysis of A.C. conductivity. Solid State Ionics 1983, 9-10, 277−282. (26) Sidebottom, D. L. Colloquium: Understanding ion motion in disordered solids from impedance spectroscopy scaling. Rev. Mod. Phys. 2009, 81, 999−1014. (27) Sidebottom, D. L. Dimensionality Dependence of the Conductivity Dispersion in Ionic Materials. Phys. Rev. Lett. 1999, 83, 983−986. (28) Jonscher, A. K. The ’universal’ dielectric response. Nature 1977, 267, 673−679. (29) Dyre, J. C.; Maass, P.; Roling, B.; Sidebottom, D. L. Fundamental questions relating to ion conduction in disordered solids. Rep. Prog. Phys. 2009, 72, 046501. (30) Faske, S.; Eckert, H.; Vogel, M. Li6 andLi7 NMR line-shape and stimulated-echo studies of lithium ionic hopping in LiPO3 glass. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 77, 104301. (31) Göbel, E.; Müller-Warmuth, W.; Olyschläger, H.; Dutz, H. 7Li NMR Spectra, Nuclear Relaxation, and Lithium Ion Motion in Alkali Silicate, Borate, and Phosphate Glasses. J. Magn. Reson. 1979, 36, 371− 387. (32) Pradel, A.; Michel-Lledos, V.; Ribes, M.; Eckert, H. Structural and electrical characterization of glasses in the system Li2S-SiSe2 by 29 Si MAS NMR and Raman spectroscopy. Solid State Ionics 1992, 53− 56, 1187−1193. (33) Ribes, M.; Barrau, B.; Souquet, J. L. Sulfide glasses: glass forming region, structure and ionic conduction of glasses in Na2S - XS2 (X = Si;Ge), Na2S-P2S5 and Li2S-GeS2 systems. J. Non-Cryst. Solids 1980, 38-39, 271−276. (34) Pradel, A.; Ribes, M. Lithium chalcogenide conductive glasses. Mater. Chem. Phys. 1989, 23, 121−142. (35) Sen, S.; Mukerji, T. Potential energy landscape of Li+ ions in lithium silicate glasses: Implications on ionic transport. J. Non-Cryst. Solids 2005, 351, 3361−3364. (36) Habasaki, J.; Leon, C.; Ngai, K. L. Dynamics of glassy, crystalline and liquid ionic conductors; Springer International Publishing: 2017. (37) Avila-Paredes, H. J.; Jain, P.; Sen, S.; Kim, S. Oxygen Transport in Sc-Doped CeO2: Cation (45Sc) NMR as a Probe of Anionic Conductivity. Chem. Mater. 2010, 22, 893−897. (38) Sen, S.; Edwards, T. G.; Kim, S. K.; Kim, S. Investigation of the Potential Energy Landscape for Vacancy Dynamics in Sc-Doped CeO2. Chem. Mater. 2014, 26, 1918−1924.

8710

DOI: 10.1021/acs.chemmater.7b02858 Chem. Mater. 2017, 29, 8704−8710