Complex hydrides have collected recent attention as a new class of solid electrolytes with potential applications in all-solid-state batteries. To improve ionic conduction in the complex hydrides, multi-cation crystal structure can be attractive. It will allow tuning the cation dynamics via structure modification depending on types and number of additional cations. However, multi-cation crystal structure struggles with the inter-cation scattering among different cations. To address this issue, understanding the conduction mechanisms in the multi-cationic crystals is indispensable. Here, we study cationic conduction in a double-cation (Li and Na) complex hydride Li3Na(NH2)4, which is formed by replacing Li (with Na) from specific lattice site of LiNH2 without altering the crystal symmetry. The nuclear magnetic resonance (NMR) measurements found that Li3Na(NH2)4 is a Li-ion conductor with negligibly small Na-ion conduction. This finding is critically important to elucidate Li-ion conduction mechanism in Li3Na(NH2)4. Enhanced Li-ion conduction in Li3Na(NH2)4 is achieved by (a) suppressing diffusion of Na cation trapped at the strategically located 2c lattice sites under deep potential well; and (b) by increasing the Li defect concentration influenced by the larger volume of the Li metastable sites due to Na substitution into LiNH2. Our study will provide the design principle for multi-cation complex hydrides, and accelerate development of superior solid electrolytes for all-solid-state batteries.

Multi-cation crystal structure with dissimilar cations in a single phase has high potential to enhance the ion conductivity of solid electrolytes, as was exemplified by almost million-times higher ion conductivity in RbAg4I5 than in AgI.1 However, to take advantage of the multi-cation crystal structure with respect to ion conduction, it is imperative to suppress the inter-cation scattering between different cations, often described in terms of mixed alkali effect (MAE).2–4 In most cases, the inter-cation scattering lowers the ion conductivity of multi-cation solids by two- to six-orders of magnitude than that of single-cation solids.5–9 Therefore, to extract potential of multi-cation crystal structure, understanding the ion conductivity mechanisms in the multi-cation solids is critical. However, within the limited number of studies,10,11 where multi-cation structures have shown higher ion conduction than its mono-cation counterparts, it remains difficult to compare exclusively the impact of the cations in these improvements largely due to very different crystal structures formed by the mono- and the multi-cationic ion conductors.

Our previous study12 suggests that double-cation complex amide Li3Na(NH2)4, which belongs to a prospective group of lightweight complex hydrides extensively investigated as hydrogen storage materials,13–15 is an ideal structure to study the mechanism of enhanced ionic conductivity in the multi-cation solids. The improvement of conductivity in this double-cation structure is confirmed by approximately three orders of magnitude larger conductivity of Li3Na(NH2)4 than that of its mono-cation counterpart LiNH2. More importantly, structural similarity between LiNH2 and Li3Na(NH2)4 (both have the same crystal structure type I4¯ in which Na cations orderly occupy 2c sites of LiNH2 to form Li3Na(NH2)4) provides a simplified situation that helps us to elucidate the impact of different cations in improving the ionic conductivity. Despite such interesting prospect, ionic conductivity enhancement mechanisms as well as the degree of possible inter-cation scattering between Li and Na was still to be assessed due to the lack of the information about individual contribution of Li and Na cation in the ionic conductivity of Li3Na(NH2)4.

This study reveals the mechanisms, which contribute to the enhancement of the conductivity in the double cation solids, by identifying first the dominant conductive cation species in Li3Na(NH2)4, and then by considering the crystal structure to analyze the conductivity data comprehensively. The solid-state NMR measurements find that the dominant conductive cation in Li3Na(NH2)4 is Li. Combining this finding with the structural information and the conductivity data, we conclude that the Na cations are trapped at the 2c sites surrounded by the deep potential well in Li3Na(NH2)4, leading to a low possibility of the inter-cation scattering between Li and Na. Further, this approach verifies that enhancement of the conductivity in Li3Na(NH2)4 is caused by the increase of the Li-defect (carrier) as a result of increased Li hopping into volume-expanded metastable interstitial sites created by Na substitution in LiNH2.

Powder Li3Na(NH2)4 was prepared by mechanical ball milling a 3:1 molar mixture of LiNH2 and NaNH2 powders as described in the literature.16,17 Conductive ion species in Li3Na(NH2)4 were determined from 7Li and 23Na NMR measurements. Electrical conductivities were evaluated using the AC and DC methods. The experimental details as well as the data from the XRD studies are given in the supplementary material.

Solid-sate NMR measurements found that the dominant ion conductive species in Li3Na(NH2)4 is Li. We qualitatively characterized the individual contributions of Li and Na in ion conduction by comparing 7Li and 23Na NMR spectra collected for Li3Na(NH2)4. Figure 1(a) and (b) show the 7Li and 23Na NMR spectra, respectively, at various temperatures in the range of room temperature (RT) to 423 K. The shape of the 7Li NMR spectra gradually changes with temperature. Below 323 K, only a broad peak is observed, while at 373 K, a central sharp peak begins to be resolved from the broad peak. The intensity of this sharp peak increases at 423 K. The appearance of the central sharp peak at elevated temperatures indicates motional narrowing caused by the movement of Li.18 Coexistence of the broad and sharp peaks observed in 7Li NMR spectra can be the indication of different Li diffusivities for different crystal sites (4f and 2a sites) in Li3Na(NH2)4.21,22 In contrast, the shape of the 23Na NMR spectra in Fig. 1(b) do not show clear temperature dependence. A nearly unchanged broad peak is observed for the 23Na NMR profiles, indicating that Na conduction is negligible in comparison to Li conduction in Li3Na(NH2)4.

FIG. 1.

NMR spectra of (a) 7Li and (b) 23Na obtained from Li3Na(NH2)4 over the temperature range of RT 423 K.

FIG. 1.

NMR spectra of (a) 7Li and (b) 23Na obtained from Li3Na(NH2)4 over the temperature range of RT 423 K.

Close modal

To support our finding that it is the Li-conduction that dominates in Li3Na(NH2)4, we characterize the electrolyte-electrode interface formed by Li3Na(NH2)4 with Li (or, Na) metal electrodes by comparing the Cole-Cole plots shown in Figure 2. The impedance plots are obtained at 323 K for disc-shaped Li3Na(NH2)4 pellets of 8 mm diameter and about 1 mm thickness formed by pressing Li3Na(NH2)4 powder (see inset of Figure 2). Shape of the impedance plots change depending on the type of the electrodes. As shown in Figure 2, when a pair of Li electrodes is used, there is only one semicircle. The capacitance value obtained by fitting the semicircle is 120 pF. This value indicates that the semicircle can be ascribed to the ion conduction in Li3Na(NH2)4 crystal. When a pair of Na electrodes is used, two semicircles appear at the higher and the lower frequency ranges. The values of the capacitance obtained by fitting these semicircles are 110 pF and 15 nF. The 110 pF capacitance should correspond to the ion conduction in the crystal of Li3Na(NH2)4 whereas the latter value should be associated with the ion conduction at grain boundaries or at electrode/electrolyte interface. Considering that we have replaced only the electrodes (Na or Li), the grain boundaries in Li3Na(NH2)4 are not likely the origin of the second semicircle. We, therefore, argue that the interface between Li3Na(NH2)4 and the Na electrodes is responsible for the second semicircle. It is very likely that, Li3Na(NH2)4 being a Li-ion conductor, the diffusing Li cations may face resistance against the Na electrode since the Li diffusion into the electrolyte-electrode interface must undergo Li-Na exchange and/or inter-cation scattering. The present analysis, thus, is consistent with the NMR data indicating Li-dominant conduction in Li3Na(NH2)4. The results have additional importance from the viewpoint of battery design in the sense that the metallic Li can be a suitable electrode for all-solid-sate batteries using Li3Na(NH2)4 as the solid electrolyte for the Li-ion battery.

FIG. 2.

Cole-cole plots obtained at 323 K using a pair of Li electrodes (solid line) and a pair of Na electrodes (dotted line) for Li3Na(NH2)4. Representative measurement frequencies are added on the plots. Inset shows picture of the sample with Li electrodes.

FIG. 2.

Cole-cole plots obtained at 323 K using a pair of Li electrodes (solid line) and a pair of Na electrodes (dotted line) for Li3Na(NH2)4. Representative measurement frequencies are added on the plots. Inset shows picture of the sample with Li electrodes.

Close modal

Experimental proof of dominant Li conduction in Li3Na(NH2)4 enables us to identify the following specific aspects of the crystal structure that effectively supress Na conduction. Figure 3 compares the crystal structures of LiNH2 and Li3Na(NH2)4, drawn from the published structural information.16,17 LiNH2 and Li3Na(NH2)4 have the same crystal structure type (I-4). The difference comes from the 2c site; this site is occupied by Li in LiNH2 whereas this is occupied by Na in Li3Na(NH2)4. The first principles calculations for LiNH2 indicated that Li atoms at the 2a and 2c sites are trapped by their surrounding deep potential wells and largely remain inactive in cation diffusion.19,20 In contrast, the potential well around Li at the 4f site is shallow and allows some motional degrees of freedom. In Li3Na(NH2)4, Li occupies the 4f and 2a sites, and Na occupies the 2c sites. From this structural comparison between LiNH2 and Li3Na(NH2)4, we argue that Na cations at 2c sites affect the potentials around Li cations at the 4f and 2a sites that allows higher Li defect concentration in Li3Na(NH2)4.

FIG. 3.

Crystal structures of LiNH2 (top) and Li3Na(NH2)4 (bottom) viewed from three different directions. Red, green, blue, and sky-blue spheres represent Li, N, H, and Na, respectively. LiNH2 is formed from three crystallographically different Li atoms located at 2a (0, 0, 0), 4f (0, 1/2, 0.009), and 2c (0, 1/2, 1/4) in space group I-4 (No. 82). Tetrahedra formed by four nitrogen atoms have lithium metastable sites (white spheres).

FIG. 3.

Crystal structures of LiNH2 (top) and Li3Na(NH2)4 (bottom) viewed from three different directions. Red, green, blue, and sky-blue spheres represent Li, N, H, and Na, respectively. LiNH2 is formed from three crystallographically different Li atoms located at 2a (0, 0, 0), 4f (0, 1/2, 0.009), and 2c (0, 1/2, 1/4) in space group I-4 (No. 82). Tetrahedra formed by four nitrogen atoms have lithium metastable sites (white spheres).

Close modal

To obtain further insight into the mechanism that improves the Li-ion conductivity in double-cation structure, we consider the ion conductivity data reported in our previous study for Li3Na(NH2)4.12 As shown in Figure 4, the conductivity of double-cation Li3Na(NH2)4 is approximately two to three orders of magnitude larger than that of mono-cationic LiNH2. Furthermore, the DC conductivity measurements (inset of Figure 4) found that ionic conduction is dominant and the electron conduction is negligible in Li3Na(NH2)4.

FIG. 4.

Electrical AC conductivity data in (i) Li3Na(NH2)4, (ii) NaNH2, and (iii) LiNH2 within the temperature range 300–423 K measured during heating and cooling. Inset: DC conductivity of Li3Na(NH2)4 measured at 343 K for 60 min using blocking (Mo) and non-blocking (Li) electrodes. Adapted with permission from B. Paik, J. Phys. Chem. C 121, 23906 (2017).12 Copyright 2017 American Chemical Society.

FIG. 4.

Electrical AC conductivity data in (i) Li3Na(NH2)4, (ii) NaNH2, and (iii) LiNH2 within the temperature range 300–423 K measured during heating and cooling. Inset: DC conductivity of Li3Na(NH2)4 measured at 343 K for 60 min using blocking (Mo) and non-blocking (Li) electrodes. Adapted with permission from B. Paik, J. Phys. Chem. C 121, 23906 (2017).12 Copyright 2017 American Chemical Society.

Close modal

Analysis of the conductivity data can show that higher Li-ion conductivity in Li3Na(NH2)4 than that of LiNH2 is mainly due to increase in the Li-defect concentration. Fitting the temperature (T) dependent AC conductivity (σ) data (Figure 4) into the equation σT = A0exp[−(Ea)/kBT], where A0 is the pre-exponential factor (also termed as the jump frequency), Ea is the activation energy for the ion conduction, and kB is the Boltzmann constant, gives us A0 7.3×1010 s-1 for Li3Na(NH2)4 and ∼ 1.0×107 s-1 for LiNH2, respectively. The jump frequency A0 is proportional to (i) concentration of the charge carrier (i.e., conducting Li ion in the present case) and, (ii) the number of the diffusion sites for Li in a unit cell. In the structures with identical symmetry the latter entity does not change. Therefore, the higher value of A0 for Li3Na(NH2)4 than that for LiNH2 should be attributed to the increase in Li-carrier concentration. In our case, the Li-carrier concentration could be increased by the number of Li-defects which are generated due to the lattice expansion from LiNH2 (a = 5.034 Å and c = 10.255 Å) to Li3Na(NH2)4 (a = 5.081 Å and 11.511 Å) as a consequence of Li substitution with Na (ionic radius of Li+: 0.59 Å, ionic radius of Na+: 0.99 Å).23 This structural modification increases the volume of the tetrahedra16,17 containing possible metastable Li sites (2b and 2d), as illustrated in Figure 3 and summarized in Table I. Such increase in the volume of the metastable Li sites will prompt Li-hopping into the metastable sites leaving Li-defects in their original sites.

TABLE I.

Volumes of the tetrahedra containing three types of possible Li metastable sites (2b, 2d, 4e) in LiNH2 and Li3Na(NH2)4.

Volume of tetrahedra (Å3)
Type of metastable siteLiNH2Li3Na(NH2)4
2b 5.79 7.45 
2d 6.33 7.48 
4e 5.47 5.40 
Volume of tetrahedra (Å3)
Type of metastable siteLiNH2Li3Na(NH2)4
2b 5.79 7.45 
2d 6.33 7.48 
4e 5.47 5.40 

Although we assumed Na to be completely immobile in the above discussion, a slightly higher Ea of Li3Na(NH2)4 (0.92 eV) than that of LiNH2 (0.87 eV) requires some modification in the model. The Ea is the sum of Ec and Ed, where Ec and Ed are, respectively, the activation energies to create Li defects and diffusion energy of the ion. Considering the carrier/defect concentration nexp[−Ec/kB·T], a higher A0 and, thereby, a higher n indicates a smaller Ec when it is compared with LiNH2. Accordingly, the Ed for Li3Na(NH2)4 must be higher than that for LiNH2 (otherwise Ea would have become smaller for Li3Na(NH2)4). If there is moderate Na-ion conduction, the higher Ed for Li3Na(NH2)4 can be explained by the Li-Na inter-cation scattering. It demands more energy to undergo a larger strain/deformation to accomplish diffusion of cations with different size (in this case, Li and Na) than that in the mono-cation (Li) diffusion process.24,25

In practice, the ion diffusion in double-cation solids should be more complicated. The diffusing Li can occupy a vacant 2c (Na vacancy) site as well as being able to replace Na from this site by, for example, the interstitialcy diffusion. This can include the 2c sites also (Na sites) in the Li diffusion paths, resulting in a somewhat mixed diffusion of Li and Na, leading to the inter-cation scattering. This mixed-cation diffusion raises the Ed because it demands more energy to undergo a larger strain/deformation to accomplish diffusion of cations with different size (in this case, Li and Na) than that in the mono-cation (Li) diffusion process24-252. It is important to recognize in this aspect that most of the Na is trapped at the 2c site, and thus the inter-cation scattering is negligible in Li3Na(NH2)4.

In summary, this study reports a novel crystal structure design that can underline an important use of the multi-cation structure in relation to the fast ion conduction. The effectiveness of the crystal structure design is elucidated by considering the crystal structure as well as by analysing the conductivity data for the double-cation amide Li3Na(NH2)4 with the aid of the knowledge that the dominant conductive species in Li3Na(NH2)4 is Li. Substitution of Li cations at the lattice sites surrounded by deep potential well with Na cation appears to be a key design-aspect to modify the crystal structure to enhance Li-ion conductivity without causing serious inter-cation scattering. In addition, choosing substituting cations with larger ionic radius is effective to enhance the ion conductivity via lattice expansion. The method should be applied to various complex hydrides including Na-conductive complex hydrides,26 and possibly, to other groups of solids. The proposed approach can also potentially be combined with state-of-the-art manipulation of anion structures for lightweight complex hydrides27–29 leading to design flexibility of fast ion conductors at room temperature. Our study, thus, can help designing high-performance solid electrolytes necessary for all-solid-state battery applications.

See supplementary material for supporting Experimental procedures and XRD profiles.

The authors gratefully acknowledge support from the Target Project 4 of WPI–AIMR, Tohoku University; Collaborative Research Center on Energy Materials, Tohoku University; JSPS KAKENHI Grant Nos. 25220911, 16K06766, 15K14155, 18H01727, 15H05546, and 18H05513; and The TEPCO Memorial Foundation. The authors thank Aidan G. Young for editing a draft of this manuscript.

The authors have no conflicts to declare.

1.
D. O.
Raleigh
,
J. Appl. Phys.
41
,
1876
(
1970
).
2.
D. E.
Day
,
J. Non-Cryst. Solids
21
,
343
(
1976
).
3.
M. D.
Ingram
,
Phys. Chem. Glasses
28
,
215
(
1987
).
4.
B.
Vessal
,
G. N.
Greaves
,
P. T.
Marten
,
A. V.
Chadwick
,
R.
Mole
, and
S.
Houde-Walter
,
Nature
356
,
504
(
1992
).
6.
J.
Swenson
and
S.
Adams
,
Phys. Rev. Lett.
90
,
155507
(
2003
).
7.
M.
Meyer
,
V.
Jaenisch
,
P.
Maass
, and
A.
Bunde
,
Phys. Rev. Letts.
76
,
2338
(
1996
).
8.
S.
Nariki
,
S.
Ito
,
K.
Uchinokura
,
T.
Uchida
, and
N.
Yoneda
,
Solid State Ionics
36
,
103
(
1989
).
9.
C. T.
Moynihan
and
A. V.
Lesikar
,
J. Amer. Ceram. Soc.
64
,
40
(
1981
).
10.
L.
He
,
H.-W.
Li
,
H.
Nakajima
,
N.
Tumanov
,
Y.
Filinchuk
,
S.-J.
Hwang
,
M.
Sharma
,
H.
Hagemann
, and
E.
Akiba
,
Chem.Mat.
27
,
5483
(
2015
).
11.
T. J.
Udovic
,
M.
Matsuo
,
W. S.
Tang
,
H.
Wu
,
V.
Stavila
,
A. V.
Soloninin
,
R. V.
Skoryunov
,
O. A.
Babanova
,
A. V.
Skripov
,
J. J.
Rush
,
A.
Unemoto
,
H.
Takamura
, and
S.
Orimo
,
Adv.Mat.
26
,
7622
(
2014
).
12.
B.
Paik
,
J. Phys. Chem. C
121
,
23906
(
2017
).
13.
S.
Orimo
,
Y.
Nakamori
,
J. R.
Eliseo
,
A.
Züttel
, and
C. M.
Jensen
,
Chem. Rev.
107
,
4111
(
2007
).
14.
Y.
Nakamori
and
S.
Orimo
,
J. Alloys. Compd.
370
,
271
(
2004
).
15.
B.
Paik
,
M.
Matsuo
,
T.
Sato
,
Li.
Qu
,
A. R.
Wolczyk
, and
S.
Orimo
,
Appl. Phys Letts.
108
,
213903
(
2016
).
16.
H.
Jacobs
and
B.
Harbrecht
,
J. Less-common Met.
85
,
87
(
1982
).
17.
R. L.
Lowton
,
M. O.
Jones
,
W. I. F.
David
,
S. R.
Johnson
,
M.
Sommariva
, and
P. P.
Edwards
,
J. Mat. Chem.
18
,
2355
(
2008
).
18.
J.
Kawamura
,
N.
Kuwata
,
Y.
Nakamura
,
T.
Erata
, and
T.
Hattori
,
Solid State Ionics
154-155
,
183
(
2002
).
19.
A.
Michigoe
,
T.
Hasegawa
,
N.
Ogita
,
T.
Ichikawa
,
Y.
Kojima
,
S.
Isobe
, and
M.
Udagawa
,
J. Phys. Soc. Jap.
81
,
94603
(
2012
).
20.
B.
Paik
,
T.
Hasegawa
,
I.
Ishii
,
A.
Michigoe
,
T.
Suzuki
,
M.
Udagawa
,
N.
Ogita
,
T.
Ichikawa
, and
Y.
Kojima
,
Appl. Phys. Letts.
100
,
151911
(
2012
).
21.
K.
Hayamizu
,
Y.
Matsuda
,
M.
Matsui
, and
N.
Imanishi
,
Solid State Nucl. Magn. Reson.
70
,
21
(
2015
).
22.
M.
Matsuo
,
S.
Kuromoto
,
T.
Sato
,
H.
Oguchi
,
H.
Takamura
, and
S.
Orimo
,
Appl. Phys. Lett.
100
,
203904
(
2012
).
23.
24.
V.
Belototsky
,
J. Non-Cryst. Solids
353
,
1078
(
2007
).
25.
V.
Belototsky
,
J. Non-Cryst. Solids
356
,
129
(
2010
).
26.
M.
Matsuo
,
T.
Sato
,
Y.
Miura
,
H.
Oguchi
,
Y.
Zhou
,
H.
Maekawa
,
H.
Takamura
, and
S.
Orimo
,
Chem. Mater.
22
,
2702
(
2010
).
27.
A.
Wolczyk
,
B.
Paik
,
T.
Sato
,
C.
Nervi
,
M.
Brighi
,
S. P.
GharibDoust
,
M.
Chierotti
,
M.
Matsuo
,
G.
Li
,
R.
Gobetto
,
T.
Jensen
,
R.
Černý
,
S.
Orimo
, and
M.
Baricco
,
J. Phys. Chem. C
121
,
11069
(
2017
).
28.
M.
Matsuo
,
A.
Remhof
,
P.
Martelli
,
R.
Caputo
,
M.
Ernst
,
Y.
Miura
,
T.
Sato
,
H.
Oguchi
,
H.
Maekawa
,
H.
Takamura
,
A.
Borgschulte
,
A.
Züttel
, and
S.
Orimo
,
J. Amer. Chem. Soc.
131
,
16389
(
2009
).
29.
M.
Matsuo
,
H.
Takamura
,
H.
Maekawa
,
H.
Li
, and
S.
Orimo
,
Appl. Phys. Lett.
94
,
084103
(
2009
).

Supplementary Material