Structure Investigations on 100LiO_1/2-(100-x)PO_5/2-xTeO₂ Fast Ionic Conducting Glasses Using Solid-State Nuclear Magnetic Resonance Spectroscopy

1.   Introduction

Phosphate glasses have been widely applied in high power laser devices ^1-3, biological materials ^4-7, and fast ionic conducting glasses ^{8, 9}. Phosphate ionic conducting glasses, which can be used in "all-solid-state" batteries, are attracting more attention. "All-solid-state" battery is considered to be a very important solution for next-generation rechargeable batteries due to the simple structure, high energy density, and great safety. The properties of phosphate ionic conducting glasses, as potential solid-state electrolyte or cathode materials, can be improved by incorporating various components due to the flexible compositions and admirable vitrification abilities ^{10, 11}. For instance, a series of Na₂O-FeO-P₂O₅ glasses were evaluated as the cathodes for sodium-ion batteries, the conductivity increased with the increase of FeO and 30Na₂O-40FeO-30P₂O₅ glass exhibited high reversible discharge capacity as 115 mAh∙g⁻¹ with a Na anode ¹². In the AgI-P₂O₅ conducting glass system, the addition of tungsten oxide (WO₃) can adjust the glass transition temperature, thermal expansion coefficient, refractive index, optical band edge, electrical conductivity, and improve considerably glass stability against water and humidity in the environment, which are important for drawing conductive fibers ^{13, 14}.

The properties of fast ionic conducting glasses are strongly related to the glass network structures. The structure investigations can help to establish the connection between the glass structure and the composition. However, further structural investigations are rarely reported. With the help of mature vibration spectroscopy technologies, such as FTIR and Raman spectra, the structures of ionic conducting glasses can be analyzed. FTIR and Raman spectroscopy was used to study the structures of solid-state glass electrolytes with the compositions of xLi₂O-(1−x)[yB₂O₃-(1−y)P₂O₅], in which more P₂O₅ could convert the [BO₃] into [BO₄] and reduce the Li⁺ ion conductivity, while more B₂O₃ could increase [BO₃] thus increasing the Li⁺ ion conductivity of glass electrolyte ¹⁵.

Recent years, advanced solid-state nuclear magnetic resonance (SSNMR) spectroscopy technologies have shown outstanding advantages on probing the structures of vitreous materials, due to their flexible and powerful capabilities on detecting the glass networks within short- and medium-range ^16-20. The structures of AgI-AgPO₃-Ag₂WO₄ ionic conducting glasses were investigated using multiple SSNMR technologies, the successive conversion from P—O—P into P—O—W linkages were observed, and the Q⁽²⁾-like chain were broken into Q⁽¹⁾ and Q⁽⁰⁾ species linked tungsten species and this structural transformation increased the glass rigidity and stability against hydrolysis reactions ¹⁹.

In this work, the structures of the glasses with the compositions of 100LiO_1/2-(100−x)PO_5/2-xTeO₂ (x = 0, 10, 20, 25, 30) are investigated using Raman spectroscopy and multiple SSNMR technologies. The evolution of phosphorus species is tracked using ³¹P MAS NMR and Raman spectra. The connectivities between phosphorus species are identified by 2D refocused INADEQUATE spectra. ¹²⁵Te WURST-QCPMG experiments are employed to probing the local chemical environments of Te atoms. The correlations between different structure units are discussed based on a random distribution model. Summarized from all discussions, a comprehensive depiction of glass networks is presented.

2.   Experimental

2.1   Sample preparation and characterization

All these glass samples with the compositions of 100LiO_1/2-(100−x)PO_5/2-xTeO₂ (x = 0, 10, 20, 25, 30) were prepared from LiPO₃ (99.9%), Li₂CO₃ (99.9%) and TeO₂ (99.99%). These raw materials were weighed according to the compositions with a total weight of 5 g, and then mixed in a platinum crucible. All these glasses were melted at 800 ℃ for 20 min and then the melts were cast on a preheated stainless-steel mold. These glass samples are labeled as 0Te, 10Te, 20Te, 25Te and 30Te for x = 0, 10, 20, 25, 30, respectively. The differential scanning calorimetry (DSC) curves were obtained using a METTLER TOLEDO TGA/DSC-1600 differential scanning calorimeter. During these measurements, glass samples were heated under N₂ atmosphere with a heating rate of 10 K∙min⁻¹. Raman spectra were obtained by a Renishaw in via Raman microscope with an excitation wavelength of 488 nm.

2.2   Solid-state NMR experiments

In this work, all NMR measurements were operated on a Bruker Avance Ⅲ HD 500 MHz spectrometer (11.7 T). All ³¹P single pulse MAS NMR spectra were obtained at 202.5 MHz using a 4 mm probe with experimental conditions as followed: the spinning rate is 12 kHz, the length of 90° pulse is 2.5 μs, the recycle delays are 320 s for all samples. And crystalline NH₄H₂PO₄ (chemical shift = 1.12 ppm) were employed to calibrate the chemical shifts of ³¹P.

To identify different ³¹P spices and obtain the correlations between these ³¹P species, two-dimensional (2D) refocused INADEQUATE experiments were adopted, in which the ³¹P species involved in P—O—P linkages can be detected by a double quantum (DQ) coherence process created based on J-coupling effect through P—O—P bond, while the isolated ³¹P species will be filtered out ^{21, 22}. Fig. 1 shows the pulse scheme of the refocused INADEQUATE and homologous coherence transfer pathway. In 2D refocused INADEQUATE spectra, the F2 dimension and the F1 dimension show the regular one quantum coherence spectrum and double quantum coherence resonance frequency, respectively. The double quantum coherence resonance frequency in the F1 dimension equals the sum of their offset frequencies and autocorrelation peaks will appear at both sides of the diagonal. In this work, 2D refocused INADEQUATE experiments were done using a 2.5 mm probe with a spinning rate of 25 kHz. The length of π/2 pulse is 2.0 μs and the recycle delay is 60 s. The DQ filtered coherence was created using an excitation sequence of 90°–τ–180°– τ– 90°, and the mixing time τ was optimized to be 3.32 ms.

Figure 1

Figure 1.  The pulse scheme of refocused INADEQUATE and homologous coherence transfer pathway.

下载: 全尺寸图片 幻灯片

The chemical environment of ¹²⁵Te nuclei was detected using the static wideband uniform-rate smooth truncation quadrupolar Carr-Purcell-Meiboom-Gill (WURST-QCPMG) technique ²³. The WURST-80 pulse shape was employed using an 8-step phase cycling. The pulses length of WURST excitation and refocusing were both 50 μs, and the excitation bandwidth was 700 kHz. To compensate the line shape distortions, which originate from transverse relaxation during the formation of frequency-dispersed echoes, the frequency was swept twice in two opposite directions and these two spectra after Fourier transformation were summed into a final spectrum ²³. The recycle delays were 100 s for all glass samples. The chemical shifts of ¹²⁵Te are calibrated by CdTe. All processing and deconvolutions of solid-state NMR spectra were done by DMFIT software package ²⁴.

3.   Results

3.1   DSC curves and Raman spectra

Fig. 2 shows the DSC curves of 100LiO_1/2-(100−x)PO_5/2-xTeO₂ (x = 0, 10, 20, 25, 30) glasses. The glass transition temperature (T_g) of these glasses is almost constant within a measurement error of 10 ℃, which is due to the little change of bridging oxygen fractions in these glasses (as discussed below). Fig. 3 shows the Raman spectra of glass 0Te, 10Te, 20Te, 25Te and 30Te. The assignments of Raman vibration bands in this work refer to the previous literature ^25-30. In glass 0Te (i.e. LiPO₃), all phosphorus species should be metaphosphate Q_0Te⁽²⁾ species, where the Q_mTe⁽ⁿ⁾ represents the phosphorus species with n bridging oxygen atoms (the oxygen atoms in P—O—P and P—O—Te linkages are both considered to be bridging oxygen atoms) and m Te atoms are connected to this [PO₄] tetrahedron. The band at 1175 cm⁻¹ corresponds to the symmetric stretching vibration of (PO2) units involving two P—O—Li linkages (v(PO2)_sym). And the shoulder at 1255 cm⁻¹ can be assigned to the asymmetric stretching vibration of (PO2) units (v(PO2)_asym). The bands at 695 cm⁻¹ are ascribed to the symmetric stretching vibration of the bridging oxygen between P—O—P linkages (v(POP)_sym) in long-chain phosphate structures, while the bands at 745 cm⁻¹ are ascribed to v(POP)_sym in short-chain phosphate structures. After TeO₂ is incorporated into glasses, a minor band at 1035 cm⁻¹ can be observed, which is the symmetric stretching vibrations of (PO3) involving three P—O—Li linkages in Q_0Te⁽¹⁾ species. The vibration bands associated with (TeO) structural units can be found at 480 and 635 cm⁻¹, which are due to the symmetric stretching vibration of Te—O—Te (v(TeOTe)_sym) and the asymmetric stretching of the continuous network composed of [TeO₄] trigonal bipyramid (tbp), respectively. The vibration band at 820 cm⁻¹ is ascribed to the [TeO₃] trigonal pyramid (tp) ^{25, 29}.

Figure 2

Figure 2.  The DSC curves of the glasses with the compositions of 100LiO_1/2-(100−x)PO_5/2-xTeO₂. The line on the top is the T_g depended on composition (x).

下载: 全尺寸图片 幻灯片

Figure 3

Figure 3.  The Raman spectra of the glasses with the compositions of 100LiO_1/2-(100−x)PO_5/2-xTeO₂. The vibration positions are marked by T1 [v(TeOTe)_sym], T2 [v(TeO₄)_asym], T3 [v(TeO₃)], P1 [v(POP)_long-chain], P2 [v(POP)_short-chain], P3 [v(PO₃)_sym], P4 [v(PO₂)_sym] and P5 [v(PO₂)_asym].

下载: 全尺寸图片 幻灯片

With the increase of TeO₂, the intensity of v(POP)_sym in long chains is gradually suppressed, while the band of v(POP)_sym in short chains is raised, indicating that long P—O—P chains are broken into short P—O—P chains and more Q_0Te⁽²⁾ species transform into Q_0Te⁽¹⁾ and Q_1Te⁽²⁾ species. Simultaneously, the transformation from Q_0Te⁽²⁾ species to Q_1Te⁽²⁾ species results in a slight broadening and shifting to lower wavenumber of v(PO2)_sym and v(PO2)_asym bands, since the symmetric and asymmetric v(PO2) in Q_1Te⁽²⁾ species have lower vibration frequency than that in Q_0Te⁽²⁾ species. Besides, when TeO₂ is added into glasses, both three- and four-coordinated Te can be observed. Te-correlated vibration bands gradually rise as TeO₂ increases.

3.2   ³¹P MAS NMR spectra

Fig. 4 shows the ³¹P MAS NMR spectra and deconvolution models of all these glasses. For glass 0Te (i.e. LiPO₃), a main peak at −22.7 ppm is observed, which is assigned to Q_0Te⁽²⁾ species. There is also a very small signal at −4.9 ppm corresponding to Q_0Te⁽¹⁾ species. This is because the excess Li₂O, due to a small number of volatilization of P₂O₅ during the melting, provides more nonbridging oxygen atoms to form Q_0Te⁽¹⁾ species. When TeO₂ is added into the glass, a new peak (at −13.6 ppm for 10Te glass) appears between the positions of Q_0Te⁽²⁾ and Q_0Te⁽¹⁾, which can be ascribed to Q_1Te⁽²⁾ species according to the chemical shift position. With the increase of TeO₂, Q_0Te⁽²⁾ species decrease significantly while Q_1Te⁽²⁾ and Q_0Te⁽¹⁾ species increase, which is consistent with the results of Raman spectra.

Figure 4

Figure 4.  (a) The ³¹P MAS NMR spectra of glasses in 100LiO_1/2- (100−x)PO_5/2-xTeO₂. The spinning sidebands are marked by asterisks. (b) The deconvolutions of ³¹P MAS NMR spectra. Dash lines represent Q_0Te⁽²⁾ (red), Q_1Te⁽²⁾ (yellow) and Q_0Te⁽¹⁾ (green) species, respectively.

下载: 全尺寸图片 幻灯片

3.3   ³¹P refocused INADEQUATE spectra

³¹P refocused INADEQUATE spectra are employed to identify the ³¹P species and detect the correlations between ³¹P species. Fig. 5 shows 2D ³¹P refocused INADEQUATE spectrum of glass 30Te. In the F2 dimension, all peaks observed in ordinary ³¹P single pulse MAS NMR spectra (see Fig. 4) can also be found, which indicates that there are no isolated ³¹P species in 30Te glass. Six connectivities Q_0Te⁽¹⁾-Q_0Te⁽¹⁾, Q_0Te⁽¹⁾-Q_1Te⁽²⁾, Q_0Te⁽¹⁾-Q_0Te⁽²⁾, Q_1Te⁽²⁾-Q_1Te⁽²⁾, Q_1Te⁽²⁾-Q_0Te⁽²⁾ and Q_0Te⁽²⁾-Q_0Te⁽²⁾ can be observed, corresponding to nine correlation peaks marked by nine translucent red dots in Fig. 5. These results indicate that all the phosphorus species are connected with each other through P—O—P bond.

Figure 5

Figure 5.  The 2D ³¹P refocused INADEQUATE spectra of glass 30Te. Nine translucent red dots in (b) represent nine autocorrelation peaks involving six possible connectivities between Q_0Te⁽²⁾ (−20.0 ppm), Q_1Te⁽²⁾ (−10.6 ppm) and Q_0Te⁽¹⁾ (−3.1 ppm) species.

下载: 全尺寸图片 幻灯片

3.4   Static ¹²⁵Te WURST-QCPMG spectra

Static ¹²⁵Te WURST-QCPMG spectra are generally preferred to probe the chemical environment of ¹²⁵Te rather than magic angle spinning due to the very wide ¹²⁵Te NMR chemical shift distributions ¹⁸. Fig. 6 illustrates the ¹²⁵Te WURST-QCPMG spectra and deconvolution models. The deconvolution parameters are list in Table 1. There are two components in each spectrum. For 10Te glass, the positions of two components are 2898 and 2262 ppm, which can be assigned to three- and four-coordinated Te ([TeO₃] and [TeO₄] species), respectively. [TeO₄] species are dominant when the concentration of TeO₂ is low, but as more PO_5/2 is substituted by TeO₂, the relative proportion of [TeO₃] gradually increases.

Figure 6

Figure 6.  The ¹²⁵Te WURST-QCPMG spectra and the dash lines show the deconvolution shapes. The whole peaks are deconvoluted into three-coordinated (green) [TeO₃] and four-coordinates (red) [TeO₄].

下载: 全尺寸图片 幻灯片

Table 1

Table 1.  The deconvolution parameters of ¹²⁵Te WURST-QCPMG spectra.

下载: 导出CSV

Sample	Unit	Position (ppm) (±5)	δcs (ppm) (±5)	ηcs	Area (%) (±5)
10Te	[TeO3]	2955	−668	0	27.7
	[TeO4]	2237	−766	0.5	72.3
20Te	[TeO3]	2968	−671	0	35.0
	[TeO4]	2382	−761	0.45	65.0
25Te	[TeO3]	2953	−671	0	37.6
	[TeO4]	2392	−791	0.45	62.4
30Te	[TeO3]	2945	−668	0	44.9
	[TeO4]	2423	−798	0.4	55.1

4.   Discussion

The deconvolution parameters of the ³¹P MAS NMR spectra (shown in Fig. 4) are summarized in Table 2. With the increase of x value, the proportion of Q_0Te⁽²⁾ species decreases while that of Q_0Te⁽¹⁾ and Q_1Te⁽²⁾ species continuously increase. All the structures of phosphorus and tellurium units are shown in Fig. 7. Both Q_0Te⁽²⁾ and Q_1Te⁽²⁾ species have one Li⁺ ion on average while Q_0Te⁽¹⁾ has two Li⁺ ions. The increase of Q_0Te⁽¹⁾ with TeO₂ indicates that Li⁺ ions prefer to stay around [PO₄] units rather than tellurium oxygen polyhedrons. However, a minor number of Li⁺ ions still interact with tellurium oxygen polyhedrons to form [TeO₃]. With the increase of TeO₂, more Li⁺ ions interact with tellurium oxygen polyhedrons and resulting in the formation of more [TeO₃].

Table 2

Table 2.  The deconvolution parameters of ³¹P MAS NMR spectra.

下载: 导出CSV

Sample	Unit	δ (ppm) (±0.5)	FWHM (ppm) (±0.5)	Area (%) (±2)
0Te	Q0Te(2)	−22.7	8.9	99.3
	Q0Te(1)	−4.9	6.2	0.7
10Te	Q0Te(2)	−22.1	9.2	65.2
	Q1Te(2)	−13.6	10.5	25.5
	Q0Te(1)	−4.1	6.5	9.3
20Te	Q0Te(2)	−20.9	9.5	36.9
	Q1Te(2)	−12.3	10.5	45.2
	Q0Te(1)	−3.8	6.5	17.9
25Te	Q0Te(2)	−20.3	9.5	25.3
	Q1Te(2)	−11.6	11.0	52.9
	Q0Te(1)	−3.5	6.7	21.8
30Te	Q0Te(2)	−20.0	11.3	16.0
	Q1Te(2)	−10.6	9.5	57.9
	Q0Te(1)	−3.1	6.8	26.2

Figure 7

Figure 7.  The structural sketches of phosphorus of Q_0Te⁽²⁾, Q_1Te⁽²⁾, Q_0Te⁽¹⁾, three- and four-coordinated Te species. The single, double and dash lines represent the single, double and noninteger bonds, respectively. Every Te atom has a pair of lone electrons and TeO_n represents n-coordinated Te.

下载: 全尺寸图片 幻灯片

The oxygen atoms in P—O—P, P—O—Te and Te—O—Te are all considered to be bridging oxygen (BO). Thus, the content of Te—BO— bond can be obtained as follows:

where F(TeO_n) is the relative fraction of TeO_n species listed in Table 1, N(Te) is the total content of Te under the stoichiometry of 100LiO_1/2- (100−x)PO_5/2-xTeO₂. Similarly, the content of P—BO— bond can be calculated as follows:

Thus, the fractions of P—BO— and Te—BO— bonds can be calculated as follows:

Here, we propose a random distribution model. We assume that all P—BO— and Te—BO— randomly bond to form P—O—P, P—O—Te and Te—O—Te linkages, then the probabilities to form P—O—P, P—O—Te and Te—O—Te linkages are:

And the total content of these three kinds of linkages is equal to the total content of BO as follows:

Thus, the theoretical contents of P—O—P, P—O—Te and Te—O—Te (under the stoichiometry of 100LiO_1/2-(100−x)PO_5/2-xTeO₂.) can be calculated according to the random distribution model as follows:

Simultaneously, the experimental contents of P—O—P, P—O—Te, Te—O—Te can be calculated from the deconvolutions of ¹²⁵Te WURST-QCPMG and ³¹P MAS NMR spectra (list in Table 1 and Table 2) as follows:

Fig. 8 shows the comparisons between the theoretical (red) and experimental (black) contents of P—O—P, P—O—Te and Te—O—Te linkages. For both P—O—P and Te—O—Te linkages, the experimental values are slightly higher than the theoretical ones. And for P—O—Te linkage, it is inverse. These indicate that P and Te atoms slightly prefer homonuclear connectivity than heteronuclear connectivity.

Figure 8

Figure 8.  Comparison between the theoretical contents calculated according to random distribution model (red) and experimental contents (black) of the P—O—P (a), P—O—Te (b) and Te—O—Te (c) linkages. All the contents are relative to the stoichiometry of 100LiO_1/2-(100−x)PO_5/2-xTeO₂.

下载: 全尺寸图片 幻灯片

In this glass system, the fractions of BO in total oxygen atoms can be calculated:

Fig. 9 shows the change trends of the total fractions of BO and the T_g values. With the increase of TeO₂, the fraction of BO and T_g have similar change trends that they almost remain constant within errors. This is not unexpected since the strength of the glass network depends on the fraction of BO. The similar BO fractions indicate the similar glass network connectivities and similar T_g in this glass system.

Figure 9

Figure 9.  The fractions of bridging oxygen atom and T_g of the glasses with the compositions of 100LiO_1/2-(100−x)PO_5/2-xTeO₂.

下载: 全尺寸图片 幻灯片

5.   Conclusions

The structures of the glasses in 100LiO_1/2-(100−x)PO_5/2-xTeO₂ (x = 0, 10, 20, 25, 30) system are investigated by solid-state NMR technologies and Raman spectroscopy. When TeO₂ is incorporated into these glasses, long P—O—P chains involved in glass networks are broken into short chains and Q_0Te⁽²⁾ species gradually transform into Q_1Te⁽²⁾ and Q_0Te⁽¹⁾ species. Q_0Te⁽²⁾, Q_1Te⁽²⁾ and Q_0Te⁽¹⁾ species are connected with each other through P—O—P. With the addition of TeO₂, a minor number of Li⁺ ions interact with tellurium oxygen polyhedrons resulting in the formation of [TeO₃]. However, Li⁺ ions prefer to stay around [PO₄] units rather than tellurium oxygen polyhedrons. Therefore, only a small fraction of [TeO₃] is formed, which increases with the content of TeO₂. Most Te atoms exist as [TeO₄] in all these glasses. With PO_5/2 being gradually replaced by TeO₂, both the T_g and the fraction of bridging oxygen are almost unchanged, which means the glass network connectivity has no obvious change. The homonuclear connectivities P—O—P and Te—O—Te show slight priority over the heteronuclear connectivity P—O—Te. In summary, this study presents a comprehensive structure study of Li-doped tellurium phosphate ionic conducting glasses. This work could promote the understanding of the glass structure dependence on compositions and the development of new ionic conducting glasses.

Acknowledgment

We thank Yujing Shen (Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences) and Sasa Yan (Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences) for their assistance on the DSC and Raman measurements, respectively.

References

1. [1]

   He, D.; Kang, S.; Zhang, L.; Chen, L.; Ding, Y.; Yin, Q.; Hu, L. High Power Laser Sci. Eng. 2017, 5, e1. doi: 10.1017/hpl.2016.46  doi: 10.1017/hpl.2016.46
   

2. [2]

   Hu, L.; Chen, S.; Tang, J.; Wang, B.; Meng, T.; Chen, W.; Wen, L.; Hu, J.; Li, S.; Xu, Y.; et al. High Power Laser Sci. Eng. 2014, 2, e1. doi: 10.1017/hpl.2014.4  doi: 10.1017/hpl.2014.4
   

3. [3]

   Campbell, J. H.; Suratwala, T. I. J. Non. Cryst. Solids 2000, 263, 318. doi: 10.1016/S0022-3093(99)00645-6  doi: 10.1016/S0022-3093(99)00645-6
   

4. [4]

   Kim, Y. P.; Lee, G. S.; Kim, J. W.; Kim, M. S.; Ahn, H. S.; Lim, J. Y.; Kim, H. W.; Son, Y. J.; Knowles, J. C.; Hyun, J. K. J. Tissue Eng. Regen. Med. 2015, 9 (3), 236. doi: 10.1002/term.1626  doi: 10.1002/term.1626
   

5. [5]

   Ahmed, I.; Lewis, M.; Olsen, I.; Knowles, J. C. Biomaterials 2004, 25 (3), 491. doi: 10.1016/S0142-9612(03)00546-5  doi: 10.1016/S0142-9612(03)00546-5
   

6. [6]

   Shah, R.; Sinanan, A. C. M.; Knowles, J. C.; Hunt, N. P.; Lewis, M. P. Biomaterials 2005, 26 (13), 1497. doi: 10.1016/j.biomaterials.2004.04.049  doi: 10.1016/j.biomaterials.2004.04.049
   

7. [7]

   Ahmed, I.; Collins, C. A.; Lewis, M. P.; Olsen, I.; Knowles, J. C. Biomaterials 2004, 25 (16), 3223. doi: 10.1016/j.biomaterials.2003.10.013  doi: 10.1016/j.biomaterials.2003.10.013
   

8. [8]

   Mangion, M.; Johari, G. P. Phys. Rev. B 1987, 36 (16), 8845. doi: 10.1103/PhysRevB.36.8845  doi: 10.1103/PhysRevB.36.8845
   

9. [9]

   Martin, S. W. J. Am. Ceram. Soc. 1991, 74 (8), 1767. doi: 10.1111/j.1151-2916.1991.tb07788.x  doi: 10.1111/j.1151-2916.1991.tb07788.x
   

10. [10]

    Brow, R. K. J. Non. Cryst. Solids 2000, 263, 1. doi: 10.1016/S0022-3093(99)00620-1  doi: 10.1016/S0022-3093(99)00620-1
    

11. [11]

    Brow, R. K.; Click, C. A.; Alam, T. M. J. Non. Cryst. Solids 2000, 274 (1–3), 9. doi: 10.1016/S0022-3093(00)00178-2  doi: 10.1016/S0022-3093(00)00178-2
    

12. [12]

    Nakata, S.; Togashi, T.; Honma, T.; Komatsu, T. J. Non. Cryst. Solids 2016, 450, 109. doi: 10.1016/j.jnoncrysol.2016.08.005  doi: 10.1016/j.jnoncrysol.2016.08.005
    

13. [13]

    Rioux, M.; Ledemi, Y.; Viens, J.; Morency, S.; Ghaffari, S. A.; Messaddeq, Y. RSC Adv. 2015, 5 (50), 40236. doi: 10.1039/C5RA00681C  doi: 10.1039/C5RA00681C
    

14. [14]

    Rioux, M.; Ledemi, Y.; Messaddeq, Y. J. Non. Cryst. Solids 2017, 459, 169. doi: 10.1016/j.jnoncrysol.2017.01.011  doi: 10.1016/j.jnoncrysol.2017.01.011
    

15. [15]

    Cho, K. I.; Lee, S. H.; Shin, D. W.; Sun, Y. K. Electrochim. Acta 2006, 52 (4), 1576. doi: 10.1016/j.electacta.2006.02.065  doi: 10.1016/j.electacta.2006.02.065
    

16. [16]

    de Oliveira, M.; Oliveira, J. S.; Kundu, S.; Machado, N. M. P.; Rodrigues, A. C. M.; Eckert, H. J. Non. Cryst. Solids 2018, 482, 14. doi: 10.1016/j.jnoncrysol.2017.11.052  doi: 10.1016/j.jnoncrysol.2017.11.052
    

17. [17]

    Larink, D.; Rinke, M. T.; Eckert, H. J. Phys. Chem. C 2015, 119 (31), 17539. doi: 10.1021/acs.jpcc.5b04074  doi: 10.1021/acs.jpcc.5b04074
    

18. [18]

    de Oliveira, M.; Amjad, R. J.; de Camargo, A. S. S.; Eckert, H. J. Phys. Chem. C 2018, 122 (41), 23698. doi: 10.1021/acs.jpcc.8b07827  doi: 10.1021/acs.jpcc.8b07827
    

19. [19]

    Blais-Roberge, M.; Santagneli, S. H.; Messaddeq, S. H.; Rioux, M.; Ledemi, Y.; Eckert, H.; Messaddeq, Y. J. Phys. Chem. C 2017, 121 (25), 13823. doi: 10.1021/acs.jpcc.7b03684  doi: 10.1021/acs.jpcc.7b03684
    

20. [20]

    Dongbei, L.; Shuai, X.; Zhiwu, Y. Chin. J. Magn. Reson. 2017, 34 (1), 115. doi: 10.11938/cjmr20170114  doi: 10.11938/cjmr20170114
    

21. [21]

    Lesage, A.; Bardet, M.; Emsley, L. J. Am. Chem. Soc. 1999, 121 (47), 10987. doi: 10.1021/ja992272b  doi: 10.1021/ja992272b
    

22. [22]

    Guerry, P.; Smith, M. E.; Brown, S. P. J. Am. Chem. Soc. 2009, 131 (33), 11861. doi: 10.1021/ja902238s  doi: 10.1021/ja902238s
    

23. [23]

    O'Dell, L. A.; Schurko, R. W. Chem. Phys. Lett. 2008, 464 (1–3), 97. doi: 10.1016/j.cplett.2008.08.095a  doi: 10.1016/j.cplett.2008.08.095a
    

24. [24]

    Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calvé, S.; Alonso, B.; Durand, J. O.; Bujoli, B.; Gan, Z.; Hoatson, G. Magn. Reson. Chem. 2002, 40 (1), 70. doi: 10.1002/mrc.984  doi: 10.1002/mrc.984
    

25. [25]

    Chowdari, B. V..; Kumari, P. P. Mater. Res. Bull. 1999, 34 (2), 327. doi: 10.1016/S0025-5408(99)00012-4  doi: 10.1016/S0025-5408(99)00012-4
    

26. [26]

    Himei, Y.; Osaka, A.; Nanba, T.; Miura, Y. J. Non. Cryst. Solids 1994, 177, 164. doi: 10.1016/0022-3093(94)90526-6  doi: 10.1016/0022-3093(94)90526-6
    

27. [27]

    Sekiya, T.; Mochida, N.; Ohtsuka, A.; Tonokawa, M. J. Non. Cryst. Solids 1992, 144, 128. doi: 10.1016/S0022-3093(05)80393-X  doi: 10.1016/S0022-3093(05)80393-X
    

28. [28]

    Irannejad, M.; Jose, G.; Jha, A.; Steenson, P. Opt. Commun. 2012, 285 (10–11), 2646. doi: 10.1016/j.optcom.2012.01.073  doi: 10.1016/j.optcom.2012.01.073
    

29. [29]

    Murugan, G. S.; Suzuki, T.; Ohishi, Y. J. Appl. Phys. 2006, 100 (2), 023107. doi: 10.1063/1.2215218  doi: 10.1063/1.2215218
    

30. [30]

    Brow, R. K.; Tallant, D. R.; Hudgens, J. J.; Martin, S. W.; Irwin, A. D. J. Non. Cryst. Solids 1994, 177, 221. doi: 10.1016/0022-3093(94)90534-7  doi: 10.1016/0022-3093(94)90534-7