Ionic Conductivity of Lithium Germanium Phosphate Glass

Subscriber access provided by University of Glasgow Library

C: Energy Conversion and Storage; Energy and Charge Transport

Ionic Conductivity of Lithium Germanium Phosphate Glass-Ceramics Luka Pavic, Kristina Sklepi#, Zeljko Skoko, Grégory Tricot, Petr Mošner, Ladislav Koudelka, and Andrea Mogus-Milankovic J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.9b03666 • Publication Date (Web): 03 Sep 2019 Downloaded from pubs.acs.org on September 3, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Ionic Conductivity of Lithium Germanium Phosphate Glass-ceramics Luka Pavić1*, Kristina Sklepić1, Željko Skoko2, Gregory Tricot3, Petr Mošner4, Ladislav Koudelka4, Andrea Moguš-Milanković1*

1Division

of Materials Chemistry, Ruđer Bošković Institute, Bijenička 54, 10000 Zagreb, Croatia

2Department

of Physics, Faculty of Science, University of Zagreb, Bijenička 32, 10000 Zagreb, Croatia

Corresponding authors:

Luka Pavić, Ruđer Bošković Institute, E-mail address: [email protected] Andrea Moguš-Milanković, Ruđer Bošković Institute, E-mail address: [email protected]

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

3LASIR

Page 2 of 63

UMR-CNRS 8516, Universite de Lille Science et Technologies, Bat C4, Cité scientifique, F-59655 Villeneuve d'Ascq Cedex, France

4Department

of General and Inorganic Chemistry, Faculty of Chemical Technology,

University of Pardubice, 53210 Pardubice, Czech Republic

KEYWORDS: Lithium germanophosphate glass-ceramics, Induced crystallization, Ionic conductivity, Impedance spectroscopy

ABSTRACT: In this study, the effect of induced crystallization on the electrical transport was studied in the mixed glass former glasses with the composition 40Li2O-(60-x)P2O5-

xGeO2, x = 0-25 mol% as potential solid electrolytes for Li-ion batteries. It has been of interest to investigate how various steps of crystallization influence electrical transport in prepared glass-ceramics. Structural properties of obtained glass-ceramics, which contain single to multi-crystalline phases, are characterized by XRD, MAS NMR and SEM and then correlated with electrical properties studied using Impedance Spectroscopy. For GeO2-free glass-ceramic a slight increase in the electrical conductivity is evidenced


2

Page 3 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


whereas a conductivity decrease for glass-ceramics containing up to 20 mol% of GeO2 is related to the reduction of a number of lithium ions in residual glassy phase since the LiPO3 crystalline phase is formed. The crystallization in the glass-ceramics with higher GeO2 content causes an increase in the electrical conductivity. This increase is a result of two simultaneous contributions. One is the formation of crystallites with well-defined shapes, which pronounces easy conduction pathways for lithium ions transport within crystalline grains and along crystalline grain boundaries. And the second one is the increase of predominantly phosphate amorphous phase for samples Ge-25.

1. Introduction

In recent years, a constant request for developing a new solid electrolyte suitable for lithium batteries applications has given rise to the investigations of numerous lithium materials. Much effort in the investigations of these materials has been focused on the improvement of the lithium-based storage materials.1-3 In parallel, accelerating interest in the lithium ions transport has made it possible to interpret trends in ionic conductivity and


3


Page 4 of 63

its dependence on the microstructure. It is well known that the nanomaterials in the glass matrix exhibit different, usually superior properties than the bulk glasses which leads to a new field of the research of glass-ceramics.4 Glass-ceramics are produced through the controlled crystallization processes of certain glass composition and show several advantages over parent glass such as thermal and chemical stability, mechanical strength and if contain nanocrystals, an optical transparency.5 Various properties, especially the ionic conductivity, depend on size, type, and distribution of crystalline phases created within the glass matrix. Therefore, one of the important features of the ionic conductivity in the glass-ceramics is that along with mobile ion concentration and composition, the microstructure strongly affects the resulting ionic transport with regards to the possible existence of easier conduction pathways. A large body of the work on the structural and electrical properties of lithium materials includes perovskite-like structures,6,7 sulfide glasses,8,9 and phosphates with NASICONlike structures.10,11 Due to the specific characteristics and ability to form a solid solution, NASICON-type glasses can be synthesized in various chemical compositions using different synthesis routes. However, well-controlled glass crystallization of NASICON-


4

Page 5 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


type glasses requires homogenous nucleation to obtain a single crystalline composition, which shows a high overall conductivity and reveals the remarkably high contribution of crystalline grains to the ionic conductivity.12-14 On the other hand, in a search for new lithium-based materials, the mixed glass former (MGF) systems such as lithium borophosphate glasses have been extensively studied due to their interesting structural and physical property changes upon network modifications.15-17 Mixed glass former effect has been also studied in various ternary and quaternary

systems

such

as

alkali

germanophosphate,

borogermanate

and

vanadophosphate glasses.18-22 In these studies, it was found that the enhancement of the conductivity is a function of glass formers content and structural modifications in the glass network. In our recent papers23-25 the introduction of germanium oxide as a third glass former, to the lithium borophosphate glasses, containing both constant and variable lithium ions concertation, causes the changes in the glass structure which results in DC conductivity increase. As an incorporation of germanium units into phosphates structure increases, the stronger cross-linkage of the glass network through the formation of a high


5


Page 6 of 63

amount of heteroatomic P-O-Ge bonds creates a favorable interaction of lithium ions dynamics and germanophosphate network leading to the maximum in DC conductivity. Further, the changes of the local structure in ternary lithium germanophosphate and quaternary lithium borogermanophosphate glasses were studied by state-of-the-art NMR technique and MIGRATION concept that allowed estimating local excursion of lithium ions and understanding their diffusive dynamics.24 Despite the existing studies about the influence of the GeO2 content and annealing processes used to produce glass-ceramics, a detailed understanding of the glass-tocrystals transformations in ternary lithium germanophosphate glasses is still lacking. In particular, the investigation of the controlled crystallization of mixed glass former systems is needed to understand both the crystallization processes and the formation of various crystalline phases along with its influence on electrical transport. Thus, this study extends our previous works on the changes in the electrical conductivity of 40Li2O-(60-x)P2O5-

xGeO2, 0x25 mol%, glasses and provides a comparison to the glass-ceramics of respective composition. Moreover, the main goal is to systematically investigate the effect of crystallization temperature on the formation of the crystalline phases for series of


6

Page 7 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glasses. In the current work, we present the effect of crystallization on the changes in the electrical conductivity as a number of various crystalline phases created in the glass matrix increases. The role of particular crystalline phases that crystallized at different temperatures can lead to the dominance or absence of blocking effects in the samples. Therefore, our attention in this investigation is focused on different electrical processes present in the prepared glass-ceramics and their contributions to the overall ionic conductivity.

2. Experimental section

2.1 Sample preparation and characterization

A detailed procedure for preparation of glasses with nominal composition 40Li2O-(60-

x)P2O5-xGeO2, 0x25 mol%, has been reported elsewhere.24 Glass transition (Tg), temperatures of the onset crystallization (Tc) and peak crystallization (Tp) were determined by differential thermal analysis (DTA) on DTA 404 PC NETZSCH instrument


7


Page 8 of 63

operating in DSC mode with 10 K min-1 heating rate and presented in Table 1 along with glass composition. Glass-ceramics in disk form were obtained by heat-treatment of each glass composition, 1mm thick disk, at two crystallization temperatures, lower T1 (between glass transition, Tg, and crystallization onset temperature, Tc) and higher T2 (around crystallization peak temperature) for 24 h. The heat treatments were performed in the high-temperature furnace Nabertherm, Nabertherm GmbH, in air. Obtained glassceramics are labeled in accordance with the composition i.e. GeO2 content and crystallization temperature. For example, Ge-10@580 is glass which contains 10 mol% of GeO2 and heat treated at 580 °C.

Table 1. Glass composition and experimental data obtained by DTA (glass transition temperature, Tg, crystallization onset temperature, Tc, crystallization peak temperature,

Tp and crystallization temperatures, T1 and T2, for 40Li2O-(60-x)P2O5-xGeO2 (x=0-25 mol%) series of glasses.


8

Page 9 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Glass composition Code

(mol%) GeO

Tg

Tc

Tp

T1

T2

(°C)

(°C)

(°C)

(oC)

(oC)

Li2O

P2O5

Ge-0

40

60

0

263

429

470

370

470

Ge-5

40

55

5

340

487

518

430

520

Ge-10

40

50

10

396

566

576

500

580

Ge-15

40

45

15

428

579

588

520

590

Ge-20

40

40

20

467

571

583

520

570

Ge-25

40

35

25

465

567

577

520

580

2

The crystalline phases of glass-ceramics were identified by powder X-ray diffraction (PXRD) using automatic Philips MPD 1820 diffractometer. In the experiment, Cu Kα radiation was used at 40 kV and 30 mA (graphite monochromator, proportional counter), in Bragg–Brentano geometry. The data were obtained at room temperature (RT) in the 2θ

angle

range

from

15

º

to

70 º, with the step of 0.02 º. Crystalline phase identification was done using the ICSD-


9


Page 10 of 63

PDF database. The amount of crystalline phases was determined by the Rietveld method using commercial software XPert High-Score (Plus). The Raman spectroscopy measurements were recorded at RT using a triple monochromator (Horiba Jobin Yvon, model T64000). An argon ion laser (Innova 400-15; Wuppertal, Germany) operating at 514.5 nm with a laser power of 200 mW was used for the excitation. A microscope objective with a magnification of ×10 focused the exciting light onto the surface of the prepared glass-ceramic disks. The 1D/2D

31P

magic angle spinning nuclear magnetic resonance (MAS-NMR)

experiments were performed at 162.9 MHz on a 9.4 T spectrometer with a 4-mm HX probe operating at spinning frequencies of 10-12.5 kHz. The 1D

31P

NMR data were

collected using 1.6 s pulse length (corresponding to a π/6 flip angle), a radiofrequency field strength of 50 kHz, 16 transients and recycle delay of 1200 s allowing for quantitative measurements. The quantitative spectra were decomposed using the dmfit software to determine the relative proportions between the different phosphate signals. In order to number and separate the different crystalline phosphate compounds present within the glass-ceramic, the radio frequency-driven dipolar recoupling (RFDR) NMR sequence was


10

Page 11 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


employed.26 The produced 2D map shows through space correlation signals between the phosphate species belonging to the same crystalline structure and allows thus determining the number of phosphate crystalline compounds present in a complex mixture.27-29 The 2400 x 180 acquisition points were recorded under rotor-synchronized conditions at a spinning frequency of 10 kHz with π pulse length of 7 s, 16 transients, a recycle delay of 60 s and a mixing time of 18 ms. The chemical shift values were referred as 0 ppm to H3PO4. The microstructure of the glass-ceramics was studied by Scanning Electron Microscopy FE-SEM JSM 7000 (JEOL, Welwyn Garden City, UK) manufactured by Oxford Instruments Ltd (Oxon, UK). Samples were not coated with an electrically conductive layer and the accelerating voltage was kept low. Electrical properties were studied by impedance spectroscopy. Samples were prepared by sputtering gold electrodes, as contacts, onto both sides of glass-ceramic disks using Sputter coater SC7620. Details about geometry (electrode area and thickness) for each sample are given in the Supporting Information, Table S1. Complex impedance was measured using an impedance analyzer (Novocontrol Alpha-AN Dielectric Spectrometer,


11


Page 12 of 63

Novocontrol Technologies GmbH & Co. KG) over a wide frequency range from 0.01 Hz to 1 MHz at temperatures between 303 and 523 K. The temperature was controlled to an accuracy of 0.2 K. Experimental data were analyzed by equivalent circuit modeling using the complex nonlinear least-square (CNLLSQ) fitting procedure. The typical complex impedance plot consists of multiple semicircles with the centers below the real axis. The equivalent circuit that represents each such depressed semicircle is a parallel combination of a resistor (R) and constant-phase element (CPE). The CPE is an empirical impedance function of the type ZCPE* = A(jω)-, where A and  are the constants. The values of the resistance obtained from the fitting procedures, R and electrode dimensions (t is sample thickness and A is electrode area) were used to calculate the DC conductivity, σDC = t/(R×A).

3. Results and discussion

3.1 Thermal characterization


12

Page 13 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


DTA curves of the 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol% glasses are shown in Figure 1. It can be seen that all glasses containing 10-25 mol% GeO2 crystallize on heating within the range of 550-650 °C. With increasing GeO2 content in the region of 1025 mol% GeO2 crystallization temperatures, Tc and T2, increase and the thermal stability of glasses gradually decreases as revealed from differences between crystallization temperature and the glass transition temperatures as reported in our previous paper.24 According to the DTA for each glass sample, the temperatures of heating treatments are determined and summarized in Table 1.


13


Page 14 of 63

Figure 1. DTA curves of 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glasses recorded at heating rate 10 K min-1. Glass transition temperatures, Tg, are determined from the inflection point, Tc represents the onset crystallization, T1 is defined as the temperature between Tg and Tc, T2 is defined as the temperature around crystallization maximum, and

Tm is melting temperature. Lines represent the temperatures of heat treatments.

3.2. Structural analysis

3.2.1 PXRD analysis

Figure 2 shows the course of the formation of various crystalline phases. For GeO2-free sample heat-treated at 470 °C the PXRD pattern exhibits a wide halo on which the diffraction lines attributed to the LiPO3 crystalline phase are observed, Figure 2(a). The PXRD patterns collected for Ge-5@430 and Ge-5@520 show well-defined diffraction lines of LiPO3 crystalline phase superimposed on a wide halo characteristic for amorphous phase. The increase of GeO2 content and a slight increase of heat-treatment temperatures promote a crystallization process by affecting shape and intensity of diffraction lines in PXRD patterns for Ge-10@500 and Ge-10@580 samples. Diffraction


14

Page 15 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


lines are sharper indicating an increase in the size of grains of LiPO3 crystalline phase and reduction of the amorphous phase, which is accompanied by the disappearance of the amorphous halo. In addition, barely detectable diffraction lines related to the Ge(P2O7) that appeared in the PXRD patterns of Ge-10 sample suggest both the beginning of the crystallization of germanate crystalline phases and crystallization of pyrophosphate phase. This result is consistent with the depolymerization of phosphate chains and formation of (P2O7)4- units observed in Raman spectra (presented in the Supporting Information) for Ge-10 sample, Figure S1.30-33


15


Page 16 of 63

Figure 2. PXRD diffraction patterns for 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glassceramics prepared by heat-treatment at various temperatures for 24 h.

Figure 2(b) shows PXRD patterns with identified crystalline phases present in Ge-15, Ge-20 and Ge-25 samples heat-treated at various temperatures for 24 h. It can be seen that the PXRD patterns for these glass ceramics become more complicated with more diffraction lines attributed to the various crystalline phases. Figure 2(b). The relative proportions between the different crystalline phases (in wt%) in glass-ceramics obtained using Rietveld analysis are listed in Table 2.

Table 2. Crystalline phase composition (wt%) of studied glass-ceramics obtained by heattreatment at different temperatures for 24 h, based upon Rietveld analysis. Note: the amount of glassy phase has not been determined and thus not been taken into account in calculating amounts of crystalline phases.

Crystalline

ICSD*

Phase

code

Ge-10@ 500

580

Ge-15@ 520

590

Ge-20@ 520

570

Ge-25@ 580


16

Page 17 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


LiPO3

638

93.0

77.6

41.7

50.1

1.8

-

-

Ge(P2O7)

74876

7.0

22.4

15.6

23.6

-

-

-

99454

-

-

12.7

8.6

41.3

26.3

19.6

LiGe2(PO4)3

69765

-

-

3.8

4.2

13.9

9.0

7.6

GeO2

637459

-

-

10.4

2.8

11.2

15.7

42.1

Li2O

108886

-

-

-

-

-

14.1

-

Li2Ge7O15

174101

-

-

9.3

5.1

12.7

16.7

12.8

Li2Ge4O9

248310

-

-

5.2

3.4

10.3

8.6

9.9

Li2Ge2O5

28178

-

-

1.3

1.9

8.1

8.6

6.9

Ge

44610

-

-

-

-

0.7

1.0

1.1

Ge(HPO4)2(H2 O)

*ICSD – Inorganic Crystal Structure Database of inorganic and related structures.

One sees from Figure 2(b) and Table 2 that, for Ge-15@520 and Ge-15@590 the dominant phases formed are LiPO3 and Ge(P2O7) along with other, mostly germanate phases. It is evident from Rietveld analysis that the relative proportions of the latest crystalline phases are relatively small. As GeO2 content increases to 20 mol%, the relative proportions of the germanate-rich crystalline phases are increased. On the other hand,


17


Page 18 of 63

LiPO3 and Ge(P2O7) crystalline phases disappear except for Ge-20@520 where the relative proportion for LiPO3 is found to be 1.8 wt%. Moreover, it can be seen in Raman spectra for samples containing from 15 to 25 mol% of GeO2, see Figure S1, that the band at 650 cm-1 increases in intensity as the P-O-P linkage represented by the band at 665 cm-1 is replaced. This suggests a formation of PO-Ge linkages as well as progressive replacement of metaphosphate, Q2, chains to pyroQ1 and orthophosphate Q0 units. Such an evolution of the Raman spectra clearly shows that the addition of GeO2 has an influence on the structure of heat-treated samples by the depolymerization of the phosphate chains and incorporation of GeO2 into glassy phosphate network leading to the formation of various phosphate and germanate crystalline phases.30-33 Further, the increase of GeO2 content to 25 mol% results in the additional changes in PXRD patterns. The most prominent diffraction lines are related to crystalline GeO2 and other germanate-rich crystalline phases. Formation of such crystalline phases for glassceramics containing >20 mol% of GeO2 is in the accordance with Raman spectra displayed in Figure S1, where the bands associated to the germanate, GeO4 and GeO6,


18

Page 19 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


units are detected. However, although the intensity of Raman bands assigned to the phosphate units decreases, it is clearly evident that the addition of GeO2 modifies glass structure causing an increase of numerous germanate crystalline phases that are created in these heat-treated samples.

3.2.2 Solid-State NMR

Study of the

31P

MAS-NMR spectra started with the experiments obtained on 40Li2O-

xGeO2-(60-x)P2O5, 0x25 mol%, glass series presented in Ref. [24]. In this paper here, Figure 3 shows the

31P

MAS-NMR spectra of the crystallized samples. The complex

spectra resulting from the superimposition of broad and narrow peaks (coming from the remaining amorphous phase and the crystalline phases) give direct evidence of the crystallization experienced by the glasses. Quantification between the narrow and broad 31P

NMR signals allow to separate the P present in crystalline compounds and in the

remaining amorphous phase. The percentages of P found in the amorphous phase are


19


Page 20 of 63

shown in Table 3. It is noteworthy that this percentage is related to the total amount of P and not to the chemical composition of the amorphous phase(s).

Figure 3.

31P

MAS-NMR spectra of 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glass-

ceramics prepared by heat-treatment at various temperatures for 24 h, accompanied by representative spectra decompositions (dotted lines).

For GeO2-free sample heat-treated at 470 °C broad signals are observed which are in good agreement with our previous analysis of the mixed network germanophosphate glasses. As expected from the formulation, the binary 40Li2O-60P2O5 presents two components at -40 and -26 ppm that can be assigned to Q3 and Q2 sites, respectively.24


20

Page 21 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


It is noteworthy that very small peaks can be noticed around -22 ppm within the broad resonances indicating that crystallization occurs.

Table 3. Proportions of 31P found in the amorphous phase.

% of P in the

GeO2 / mol%

Tcryst / °C

0

470

99

5

430

97

5

520

90

10

500

35

10

580

16

15

520

14

15

590

9

20

570

7

25

520

41

25

580

23

amorphous phase

With the introduction of 5 mol% of GeO2, the phosphate network for Ge-5 sample is significantly modified with the appearance of a third component at -35 ppm that is


21


Page 22 of 63

assigned to a P connected to Ge4+ ions (see Figure 3 decomposition of Ge-5@520). This assignment, indicating that phosphate and germanate units mixed together, is in a good agreement with data reported by Behrends et al. within the study of alkali germanophosphate glasses.34 On the other hand, for the Ge-5@430 and Ge-5@520 samples, the narrow peaks on the broad signals appear suggesting pronounced crystallization processes. However, the presence of broad signals implies a relatively high percentage of P in the amorphous phase, 90 % for the Ge-5 samples. This is in good agreement with the obtained PXRD results for Ge-5@430 and Ge-5@520 where, in addition to the diffraction lines, a wide amorphous halo is observed, Figure 2(a). With further introduction of GeO2, the

31P

MAS-NMR spectra for heat-treated Ge-10,

Ge-15 and Ge-20 samples exhibit intensive narrow peaks in the -15 / -37 ppm region which indicates a progressive crystallization. In case of the Ge-10 sample, narrow signals are observed in the -24 ppm region (-22.4, -24.6, -25.3 and -26.3 ppm) characteristic of Q2 species, suggesting the presence of crystalline lithium metaphosphate in the materials. In addition, signals at -31 and -37 ppm can also be observed on the 1D

31P

MAS-NMR analysis. These two latter signals are also present in the Ge-15 sample


22

Page 23 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


accompanied by the resonance at -24.6 and -25.3 ppm. Four new signals are also observed in similar region at -15, -16.2, -22 and -26.7 ppm. More detailed analysis of this complex 1D

31P

MAS-NMR spectrum is obtained with the help of the 2D

31P

correlation

map. The 2D RFDR spectrum obtained on the Ge-15@520, Figure 4(a), shows two groups of correlated signals: a first group composed of the 4 new resonances (at -15, 16.2, -22 and -26.7 ppm) and a second one composed of the -31 and -37 ppm signals, Figure 4(b). This result indicates that different phosphate phases are present within the complex mixture, these crystalline structures containing respectively four and two phosphate sites, Figure 4(b). The resonance group of two signals at -31 and -37 ppm is assigned to crystalline β-Ge(P2O7), as observed by Losilla et al.35, as dominant crystalline phase whereas the four resonances group could also correspond to a lithium metaphosphate phase or more likely is a NMR signature of one of the other phosphate crystalline phases that are present in this glass-ceramic sample. It is noteworthy that the sample used for the 2D map acquisition is slightly different from the one analyzed by 1D NMR. Indeed, the two peaks observed around -25 ppm in Figure 3 are present in lower intensity in the sample used to acquire the 2D map. However, it shows that these two


23


Page 24 of 63

signals do not belong to the two phases identified on the 2D map. The four resonance group, Figure 3, is the only remaining signals in the Ge-20 sample and P(am) decreases reaching the lowest value (7.0%) for Ge-20@570 sample. Finally, the Ge-25 sample spectrum shows broader resonances (suggesting that the crystallinity of these compounds is lower) accompanied by two signals around -5 ppm. For glass-ceramics Ge25, as much as 23% of P is found in the amorphous phase.

ppm (a)

(b)

-35 -30 -25 -20 -20

-15 -10 -10

-20

-30 31P

Figure 4. 2D

31P

-30

ppm

ppm

chemical shift / ppm

RFDR map (a) and

31P

horizontal projections showing the presence of

two different phosphate phases (b) in the crystallized glass Ge15@520.


24

Page 25 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Now our analysis should be turned to the results of 31P MAS-NMR experiments obtained for germanophosphate glasses, which are described in detail in Refs. [23,24]. It was found that with the increase of GeO2 content in 40Li2O-xGeO2-(60-x)P2O5, 0x25 mol%, glasses the fraction number of P-O-Ge bonds continuously increase. This result indicates that the glass networks adopt a mixed character without any segregation between the phosphate and germanate networks. Furthermore, the structural parameters calculated from the decomposition of 31P MAS-NMR spectra show that at the high GeO2 content (25 mol%) the average number of P ions connected to the Ge ions is reduced, suggesting that Ge ions may form Ge-O-Ge bonds. Therefore, the formation of Ge-O-Ge bonds that occurs in Ge-25 glasses supports the idea that the number of P-O-Ge bonds becomes slightly lower which confirms a strong effect of the GeO2 insertion into the glass network. The observed structural changes are in accordance with the PXRD, Figure 2, and Raman results, Figure S1, where the formation of various germanate crystalline phases is present at higher GeO2 content (GeO2  20 mol%) accompanied with presence of Raman bands associated to the germanate GeO4 and GeO6 units, rather than an increase of the overall amorphous phase.


25


Page 26 of 63

3.2.3 SEM analysis

For a further understanding of the glass-ceramics structural changes, the insight in the microstructural features of glass-ceramics, grains and grain boundaries, was gained using Scanning Electron Microscopy (SEM). Figure 5(a) exhibits SEM micrograph of Ge-0@470. As can be seen, GeO2-free glass heat-treated at 470 °C is almost fully amorphous which is consistent with the PXRD and 31P

MAS-NMR data. The heating of Ge-5 sample at 520 °C resulted at the beginning of

the formation of crystalline grains inside of the glassy phase, Figure 5(b). The increase of GeO2 content in Ge-10@500 sample promotes the crystallization process and the first crystalline phase, LiPO3, was formed Figure 5(c). Along with the well-defined rectangular LiPO3 crystalline grains, the second small crystalline grains of Ge(P2O7) surround the LiPO3 crystallites and appeared embedded in a glassy matrix. According to Rietveld analysis the relative proportions of LiPO3 and Ge(P2O7) phases in this sample, are about 77.6 wt% and 22.4 wt%, respectively. It should be noted that using Rietveld analysis only


26

Page 27 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


relative proportions of crystalline phases are calculated since for such complicated glassceramic systems, which contain numerous crystalline phases, the residual amorphous phase cannot be determined. However, the SEM micrograph in Figure 5(c) clearly reveals that some amount of glass matrix is still present. Further evolution of the microstructure is observed for Ge-15@590 sample showing various morphologies of crystalline grains which do not have regular shapes, Figure 5(d). Therefore, it was assumed that a large number of crystalline phases identified in this sample form agglomerates of crystallites surrounded with some fraction of residual amorphous phase.


27


Page 28 of 63

Figure 5. SEM micrographs of selected 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glassceramics prepared by heat-treatment at various temperatures for 24 h.

With increasing GeO2 content and heat-treatment temperature, certain crystalline grains grow to reach the size of several m in diameter as is evidenced in Figures 5(e,f). As expected, the better defined crystal-like shape can be distinguished just for few crystalline grains.

4. Electrical properties 4.1 Complex impedance spectra For the analysis of the effect of crystallization processes in the prepared glassceramics, the results of the impedance measurements are presented by complex impedance spectra, as imaginary impedance Zʺ against real impedance Z part, Figures 6 and 7. It is known that the shape of the impedance spectra is correlated with the nature of the electrical transport processes in the samples, thus, impedance data not necessary exhibit


28

Page 29 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a single impedance semicircle. In particular, glass-ceramics very often exhibit multiple semicircles related to the crystalline grains and grain boundaries. In order to properly identify and separate different regions within the electrode-material system and determine their electrical properties,36,37 the impedance data were fitted with the equivalent circuit model comprising multiple parallel combinations of the resistor (R) and the constantphase element (CPE), each being associated to one impedance semicircle. It is worth to note that the experimental data show overlapped and depressed semicircles, especially for such complicated systems with various crystalline phases and hence their separation is limited. However, for the samples with a lower number of crystalline phases and defined crystalline grains, semicircles can be identified and separated. The use of CPE rather than a capacitor in the equivalent circuits is due depressed semicircles with center of semicircles positioned bellow the x-axis. The fitting parameters of the equivalent circuits were obtained by means of CNLLSQ fitting directly to the measured impedance data and are given in Table 4 for selected samples. It should be mentioned that for these samples the experimental data are in good accordance with the theoretical curves obtained by the fitting procedure.


29


Page 30 of 63

Table 4. The fitting parameters obtained from equivalent circuit modeling of complex impedance spectra measured at 90 °C for Ge-5 glass and selected glass-ceramics samples prepared by heat-treatment of selected starting glasses at various temperatures for 24 h.

Samples/

‡

Ge-5 Ge-5@520

Ge-10@500

5.29×106

7.54×107

3.76×108

A1§ /(sa –1)

1.97×10-11

7.43×10-12

4.34×10-12

a 1†

0.88

0.87

0.93

C 1* / F

5.64×10-12

2.43×10-12

2.68×10-12

R 2‡ / 



9.13×107

5.68×108

1.81×109

A2§ /(–1 sa)



7.97×10-10

1.33×10-11

5.34×10-11

a 2†



0.65

0.89

0.72

C 2* / F



1.94×10-10

7.27×10-12

2.15×10-11

R 3‡ / 



6.08×108

1.17×109

A3§ /(–1 sa)



1.05×10-10

7.67×10-10

a 3†



0.79

0.56

C 3* / F A4§ /(–1 sa) a 4†



5.05×10-11

7.04×10-10

Parameters

starting

R 1‡ / 

Ge-10@580

glass

1.73×10-6

5.49×10-8

1.20×10-7

4.92×10-8

0.74

0.30

0.87

1

Individual value of resistance (R) for every R-CPE circuit element from proposed

model.


30

Page 31 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


A constant (CPE capacitance) from the empirical impedance function: 𝑍 ∗ CPE =

§

1 𝐴(𝑖𝜔)𝛼. †

a constant (0  𝛼  1) from the empirical function: 𝑍 ∗ CPE = 1 𝐴(𝑖𝜔)𝛼. Unit is (–1 sa)

which means when 𝛼 = 1 (true capacitor), unit is –1 s = F. *

Individual value of capacitance (C) for every R-CPE circuit element from proposed model calculated from the equation: 𝐶 = 𝐴(𝜔max)𝛼 ― 1 [38].

Complex impedance spectra of Ge-5 starting glass measured at 90 °C and its corresponding electrical circuit is shown in Figure 6(a). The spectra consist of a single semicircle that corresponds to the bulk conduction and low-frequency spur related to the lithium ions being blocked at gold electrodes. The x-intercept represents the value of the DC resistance (R) of samples observed at a certain temperature.


31

1

2

3

4

250

CPEel

R

150

Rgl

Rbc

CPEgl

CPEbc

CPEel

100 50 0

5

6

7

0

8

50

100

Z' / M

150

200

250

Z' / M 3

1.5

Page 32 of 63

(b) Ge-5@520

200

-Z / M

8 7 (a) LiG-5, starting glass 6 Rgl 5 CPEgl 4 3 2 1 0 0

(c) Ge-10@500

(d) Ge-10@580

2 1.0

Rgl

Rg

CPEgl

0.5

CPEg

Rgb CPEgb

CPEel

-Z / G

-Z / G

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

-Z / M


0.0

1

Rg

Rgb

CPEg

CPEgb

CPEel

0 0.0

0.5

1.0 Z' / G

1.5

0

1

2

3

Z' / G

Figure 6. Complex impedance spectra measured at 90 °C for (a) Ge-5 starting glass, (b) Ge-5 heat-treated at 520 °C, (c-d) Ge-10 heat-treated at 500 °C and 580 °C respectively. Symbols (colored circles) denote experimental values, whereas solid black line corresponds to the best fit. The corresponding equivalent circuit model, comprising of multiple parallel combinations of the resistor (R) and the constant-phase element (CPE), used for fitting the data of individual sample and its interpretation is shown in each figure (defined as follows: gl-glassy phase, g-grain, gb-grain boundary, bc-beginning of crystallization and el-electrode polarization).


32

Page 33 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6(b) shows the impedance spectrum for Ge-5@520 sample. This complex impedance spectrum consists of two deformed semicircles related to two different and electrically distinct processes and an indication of spur related to the electrode polarization. Semicircles are separated using equivalent circuit modeling and appropriate model composed of two R-CPE elements and one CPE connected in series. Here, it should be noted that the capacitance values (C) are directly connected to the volume fraction of each region present in the samples, which means that the small capacitance value corresponds to the large volume fraction while large capacitance values are related to the lower volume fraction.36,37 Therefore, the obtained capacitance (C1) for the highfrequency semicircle is in the range of  10-12 F and is connected to the large bulk fraction, which for this sample is a glassy matrix. This is in good accordance with the

31P

MAS-

NMR data displayed in Table 3 where the percentage of P in the amorphous phase is found to be 90%. In addition, SEM micrograph in Figure 5(b) exhibits just the beginning of the crystallization process and structural modifications within the glassy phase. Thus, the low-frequency semicircle described by Rbc-CPEbc circuit and determined capacitance


33


Page 34 of 63

value (C2) of 1.94×10-10 is related to this second process in Ge-5@520 sample, beginning of crystallization, which is also confirmed by PXRD, see Figure 2(a). Going further in the interpretation of the Ge-10@500 and Ge-10@580 glass-ceramics, Figure 6(c,d), it is observed that the impedance spectra are more deformed which correlates well with the microstructural modifications i.e. formation of well-defined LiPO3 crystallites within glass matrix as shown in Figure 5(c). Impedance spectrum of Ge10@500 consists of three semicircles, which can be approximated by three R-CPE elements in series. High-frequency Rgl-CPEgl circuit can be ascribed to the glassy phase whereas Rg-CPEg circuit is attributed to the crystalline grains. The low-frequency RgbCPEgb circuit is associated with the grain boundaries. In the case of Ge-10@580 glassceramics heat-treated at higher temperature, the impedance spectrum shows different features as crystallization process progress due to an increase in the amount of crystalline phases embedded in amorphous matrix accompanied by microstructural changes. For this particular sample, the semicircle related to the glassy phase disappears from the complex impedance spectrum (outside of the frequency window) indicating a small fraction of the glassy phase. Therefore, the two semicircles described by Rg-CPEg and


34

Page 35 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Rgb-CPEgb circuits represent the crystalline grains and grain boundaries, respectively. Such a description agrees well with values of the fitting parameters listed in Table 4, which reveal that the highest capacitance (C1) of 2.68×10-12 F obtained just for Ge-10@500 glass-ceramic corresponds to the glassy phase. On the other hand, the fitting results of the grain capacitance (C2) for Ge-10@500 and Ge-10@580 glass-ceramics equals to 7.27×10-12 and 2.12×10-11 F, respectively. These results of fitting are consistent with structural analysis. The dominant fraction, calculated by Rietveld belongs to the LiPO3 with a value of 93.0 wt% in Ge-10@500 sample. According to PXRD a small fraction of GeP2O7 crystalline phase is identified having a relative proportion of 7.0 wt%. For glass-ceramics heat treated at a higher temperature, Ge-10@580, the relative proportion was found to be 77.6 and 22.4 wt% for LiPO3 and GeP2O7 crystalline phases, respectively. In addition, the percentage of P in the amorphous phase obtained by

31P

MAS-NMR data decreases from 35% to 16% for Ge-

10@500 and Ge-10@580 glass-ceramics confirming that the most of the phosphates in these samples are in the form of LiPO3 crystalline grains. Therefore, an increase in the relative proportion of LiPO3 crystalline phase for Ge-10 samples results in raising the


35


Page 36 of 63

overall resistance as the grain and grains boundaries are identified. Moreover, at the same time, the glassy matrix is impoverished with lithium ions, which additionally increases a sample resistance. Up to this point, the contributions of crystalline grains and grain boundaries to the overall resistance of glass-ceramics studied in this paper could be successfully determined using equivalent circuit modeling since only two crystalline phases were identified for Ge-5 and Ge-10 samples. In the further analysis of the complicated Ge-15, Ge-20 and Ge-25 glassceramic systems, which contain numerous crystalline phases, the separation of the mostly overlapped semicircles is limited. Thus, the next step in our investigation is to determine the total resistance for each glass-ceramic without attempting to depict any particular contribution. In Figure 7, Ge-15@590, Ge-20@570 and Ge-25@580 samples exhibit similar impedance spectra except for Ge-25@580 sample that shows a spur related to the electrode polarization. Total resistance, Rtot, for these three samples was determined at low-frequency intercept on the real axis.


36

Page 37 of 63

80

3 (a) Ge-15@590

400 (b) Ge-20@570

(c) Ge-25@580

60

1

300

-Z / M

-Z / G

2

-Z / G

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


40 20

200 100

0

0 0

1

2

3

Z' / G

0 0

20

40

60

80

Z' / G

0

100

200

300

400

Z' / M

Figure 7. Complex impedance spectra at 90 °C for glass-ceramics prepared by heattreatment for 24 h at various temperatures: (a) Ge-15 at 590 °C, (b) Ge-20 at 570 °C, and (e-c) Ge-25 at 580 °C.

4.2 Electrical conductivity

The values of Rtot calculated as a sum of resistances obtained from the equivalent circuit modeling were used to determine the total DC conductivity for all glass-ceramic samples. It is important to note that the DC values are identical to the values resolved from the low frequency plateau of conductivity spectra, Figure S2. The DC conductivity of all studied samples exhibits an Arrhenius temperature dependence, and hence has characteristic activation energy. The activation energy for DC conductivity, EDC, for each sample was


37


Page 38 of 63

determined from the slope of log(DCT) vs. 1000/T using equation DCT = 0*exp (-EDC / kBT) where DC is the total conductivity, 𝜎0∗ is the pre-exponential factor, EDC is the activation energy, kB is the Boltzmann constant and T is the temperature (K), see Figure S3. Obtained electrical parameters along with DC conductivity for all glass-ceramics prepared by heat-treatment at various temperatures for 24 h are shown in Table 5.

Table 5. DC conductivity, σDC, activation energy, EDC, and pre-exponential factor, σ0*, for all investigated glass-ceramics.

Glass

σDCa / (Ω cm)-1 EDC / kJ mol-1 ± 0.5%

± 0.5%

log σ0* / (Ω cm)-1 K ± 0.5%

T1-lower crystallization temperature Ge-0@370

1.7×10-7

95.8

9.31

Ge-5@430b

-

-

Ge-10@500

1.4×10-9

75.0

4.50

Ge-15@520

3.1×10-10

81.3

4.76

Ge-20@520

2.8×10-11

86.9

4.56

Ge-25@520

3.9×10-9

75.5

5.06


38

Page 39 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


T2-higher crystallization temperature Ge-0@470

4.27×10-8

77.0

6.11

Ge-5@520

1.12×10-8

82.5

6.50

Ge-10@580

4.57×10-10

94.1

6.86

Ge-15@590

5.13×10-10

95.0

7.00

Ge-20@570

1.91×10-11

87.5

4.50

Ge-25@580

4.47×10-9

72.0

4.61

a

values at 363 K

b

after heat-treatment the Ge-5@430 sample was extremely fragile and breakable, so not

measurable

In Figure 8 the dependence of DC on GeO2 content for starting glasses and corresponding glass-ceramics along with EDC is plotted showing changes in the conductivity of few orders of magnitude. The influence of the heat-treatment on DC is in good correlation with the modifications in the microstructures of glass-ceramics. In this regards, the changes in DC can be divided into three compositional regions. As can be seen from Figure 8 heat-treated GeO2-free sample, either at T1 or T2, exhibits almost two orders of magnitude higher DC conductivity than that for the starting glass, with the jump being higher for lower T1 temperature. Similar behavior is observed above


39


Tg of lithium metaphosphate glass during initial stage of crystallization.39,40 The effect of conductivity increase is a result of the formation of interfacial region between deformed LiPO3 crystalline grains and glass matrix, which lead to the creation of ionically conductive pathways for Li+ ion transport along defective glass–crystallites interfaces.

-4

starting glass

T2- higher cryst. temperature

T1- lower cryst. temperature

EDC for T2 samples

120

-5

-1

log(DC / (cm) )

-6 100

-7 -8

80

-9

EDC (kJ / mol)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 40 of 63

-10 60

-11 -12

0

5

10

15

20

25

x GeO2 / mol%

Figure 8. Compositional dependence of DC conductivity, DC, at 363 K for starting glasses 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, and glass-ceramics prepared by heat-


40

Page 41 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


treatment at various temperatures, T1 and T2, for 24 h. Dependence of activation energy,

EDC, upon composition is presented for glass-ceramics prepared by heat-treatment at T2.

For Ge-5@520 and Ge-10@580 glass-ceramics, the DC conductivity decreases as the dominant LiPO3 crystalline phase is formed reaching the relative proportion of 77.6 wt% in Ge-10@580 glass-ceramics, Table 2. As the crystallization progresses and the proportion of LiPO3 crystalline phase increases the conductivity decreases since more Li+ ions leave the glassy matrix and enter into the crystallites. A similar effect of the crystallization of LiPO3 on the electrical properties of lithium phosphate glasses was discussed in Ref. [39]. The authors reported that the DC conductivity of 50Li2O-50P2O5 (mol%) glass is almost four orders of magnitude higher than that of 50Li2O-50P2O5 glass-ceramic due to the crystallization of LiPO3. Moreover, it should be pointed out that they also observed anomalously increase in conductivity at the beginning of crystallization process, however smaller (0.3 on log scale) that in our case, which was attributed to the highly defective ionically conductive interfacial regions. It is well known that an open structure and randomly dispersed lithium ions enhance the ions transport in phosphate glasses, which leads to high ionic


41


Page 42 of 63

conductivity. On the other hand, in polycrystalline LiPO3, lithium ions move through interstitial positions between phosphate tetrahedral, which reduce their mobility and causes a drastic decrease in ionic conductivity as crystallization progresses.41 Clearly, the formation of LiPO3 is responsible for the structural changes that hamper the movement of lithium ions in LiPO3 structure. This is also consistent with observed decrease of approximately three orders of magnitude in DC conductivity for Ge-10@580 sample, Figure 8. For glass-ceramics obtained by heat-treatment at T2 of starting glass containing between 10 and 15 mol% of GeO2, the DC conductivity is almost identical. It should be kept in mind that in this compositional region along with the dominant LiPO3 crystalline phases the appearance of various germanate phases occurs which consequently keeps the DC conductivity constant. With further increase of GeO2 content to 20 mol %, the DC conductivity decreases whereas for Ge-25 glass-ceramic the DC conductivity increases for two orders of magnitude reaching the value of 4.47×10-9 (Ω cm)-1. According to EDS data, shown in Figure S4, it is likely that the large crystals correspond to the GeO2. This is consistent with Rietveld analysis as listed in Table 2. For Ge-25 glass-ceramic, the


42

Page 43 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


formation of large and well-defined crystals of  2 m as can be seen in Figure 5(f) as well as in Figure S4 and a probably better connection between other crystalline phases causes the increase of the mobility of lithium ions resulting in enhancement of DC conductivity. Thus, it seems that an increase in GeO2 content to 25 mol% and the formation of large crystalline grains increase the total ionic conductivity. However, it should be noted that the amount of phosphorus found in the amorphous phase increases, as found by

31P

MAS-NMR experiments and listed in Table 3, which provides an easier

mobility of Li+ ions along well-defined crystals. Such a behavior is in good correlation with observed spur in Ge-25@580 sample, Figure 7 (c). Therefore, this finding seeks for further investigation of glass-ceramics containing >25 mol% GeO2 expecting an additional increase in electrical conductivity. It is worth mentioning that the glasses containing ≥ 30 mol% of GeO2, prepared by the melt-quenching method, are partially crystallized, as can be seen in Figure S5. Since the intention in this paper was to investigate the effect of induced crystallization on glasses (from amorphous to crystalline) the partially crystallized Ge-samples are excluded from the present investigation and will be a matter of our next paper. To corroborate this argument, as an illustration, our preliminary result shows an


43


Page 44 of 63

increase in the DC conductivity for Ge-30 sample to 7.79×10-9 (Ω cm)-1. However, it should be kept in mind that the heat-treatments for partially crystallized samples were different than that of Ge-25@580 glass-ceramic. The presence of a small amount of crystalline germanium detected in the glassceramics with 20 and 25 mol% of GeO2 could potentially contribute to the total conductivity by introducing electronic transport. However, our DC conductivity experiments (see Figure S6) verifies that there is no contribution of electronic conductivity in these samples. Thus, the influence of crystalline Ge on the electrical properties of these materials is insignificant. Further, analysis of activation energy, EDC, reveals that the opposite trend to the DC conductivity is obeyed. However, it would be expected that the EDC for Ge-20@570 glassceramics increase similarly as DC decreases implying that the changes in microstructure have an impact on the activation energy. Thus, another aspect observed in Table 5 is that the dependence of 0* on the composition of glass-ceramics prepared at T2, shows a similar decrease as DC for glass-ceramics containing between 15 and 20 mol% of GeO2. Such behavior suggests that the microscopic parameters involved in σ0* affects the


44

Page 45 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


behavior of EDC since in this compositional region the structural modifications are influenced by the appearance of many germanate phases along with a disappearance of LiPO3 and GeP2O7 crystalline phases. Detailed analysis of EDC behavior is a subject of investigation in our next paper.

5. Conclusion

The effect of heat treatments for 24 hours on the structural changes of series of 40Li2O(60-x)P2O5-xGeO2, 5x25 mol%, glasses exhibits the evolution from glass to glassceramics. It was found that the incorporation of GeO2 plays an important role in the number and types of crystalline phases formed during crystallization processes. The strong dependence of the electrical properties of created glass-ceramics on their compositions is directly observed in the complex impedance plots. The obtained fitting results are divided into three compositional regions. In the first one, the GeO2-free glassceramic shows an increase in the electrical conductivity as a result of an early stage of structural reorganization. The second one is related to the region from 5 to 10 mol% of


45


Page 46 of 63

GeO2 where the impedance modeling demonstrates that the equivalent circuits employed here can provide a separation of each contribution to the total electrical conductivity. In this compositional region, crystallization occurs resulting in the formation of dominant LiPO3 crystalline phase, which is a major component for a decrease in DC conductivity. The third compositional region, which contains from 15 to 25 mol% of GeO2, reveals the formation of the numerous crystalline phases disallowing separation and determination of any particular contribution to the overall electrical conductivity. On the other hand, the formation of predominantly germanate crystalline phases in this multi-phased glassceramic systems leads to the creation of the favorable conduction pathways for lithium ion transport. The formation of these crystalline phases in combination with the larger amount of phosphates in the residual amorphous phase favors the Li+ ions mobility and increases DC conductivity of Ge-25 glass-ceramics. This study shows that the heat treatments performed on the simple ternary germanate glasses can produce complicated multi-phased systems. However, for the investigation of electrical conductivity of these multi-phased glass-ceramics the impedance spectroscopy, enables to clarify the electrical conductivity changes in the correlation with


46

Page 47 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the modifications of glass-ceramics structures. Such an analysis of every step in trends of the ionic conductivity of glass-ceramics is important for any further investigation of lithium-based materials, which includes mixed glass formers as promising solid electrolytes for lithium batteries.

ASSOCIATED CONTENT

Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: xxx. Detail analysis of Raman spectra for all glass-ceramics prepared by heat-treatment, details about samples geometry (thickness and electrode diameter) for each sample used for electrical characterization by means of Impedance spectroscopy together with additional details of SI results (conductivity spectra and Arrhenius plots) are provided in SI. In addition, the SI contains additional data on EDS analysis for Ge-25@580 sample, explanation on why glasses with above >25 mol% GeO2 were excluded from this research


47


Page 48 of 63

and details on additional TSC experiments performed for determining the electronic/ionic contributions to the total conductivity.

ACKNOWLEDGMENTS

The authors are very grateful to the Croatian Science Foundation for its support of this work under project no.: IP-09-2014-5863. The Czech authors are grateful for the financial support from the project No. 18-01976S of the Grant Agency of the Czech Republic. The authors would like to thank Dr. A. Šantić for the critical reading of the paper.


48

Page 49 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


REFERENCES

[1]

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, 1– 10.

[2]

Minami, T.; Hayashi, A.; Tatsumisago, M. Recent progress of glass and glassceramics as solid electrolytes for lithium secondary batteries, Solid State Ionics 2006, 117, 2715–2720.

[3]

Duclot,

M.;

Souquet,

J.

L.

Glassy

Materials

for

Lithium

Batteries:

Electrochemical Properties and Devices Performances, J. Power Sources 2001,

97-98, 610–615,

[4]

McMillan, P.W. Glass-Ceramics, in: Non-Metallic Solids: A series of

monographs, J.P. Roberts, P. Popper (Eds.), Vol 1, Academic Press, New York, 1964.


49


[5]

Page 50 of 63

Holland, W.; Beall, G.H.; Glass-ceramic technology, second ed., John Wiley & Sons, New York, 2012.

[6]

Bohnke, O. The fast lithium-ion conducting oxides Li3xLa2/3-xTiO3 from fundamentals to application, Solid State Ionics 2008, 179, 9–15.

[7]

Kim, D.-H.; Kim, D.-H.; Jeong, Y.-C.; Seo, H.-I.; Kim, Y.-C. Lithium ion migration pathways in Li3xLa2/3−x□1/3−2xTiO3, Ceram. Int. 2012, 38, S467–S470.

[8]

Tatsumisago, M.; Hayashi, A. Superionic glasses and glass–ceramics in the Li2S–P2S5 system for all-solid-state lithium secondary batteries, Solid State

Ionics 2012, 225, 342–345.

[9]

Minami, T.; Hayashi, A.; Tatsumisago, M. Recent progress of glass and glassceramics as solid electrolytes for lithium secondary batteries, Solid State Ionics 2006, 177, 2715–2720.


50

Page 51 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[10]

Narváez-Semanate,

J.L.;

Rodrigues,

A.C.M.

Microstructure

and

ionic

conductivity of Li1+xAlxTi2-x(PO4)3 NASICON glass-ceramics, Solid State Ionics 2010, 181, 1197–1204.

[11]

Yan, B.; Zhu, Y.; Pan, F.; Liu, J.; Lu, L. Li1.5Al0.5Ge1.5(PO4)3 Li-ion conductor prepared by melt-quench and low temperature pressing, Solid State Ionics 2015, 278, 65–68.

[12]

Zanotto, E.D. A bright future for glass-ceramics, Am. Ceram. Soc. Bull. 2010,

89, 19–27.

[13]

Cruz, A.M.; Ferreira, E. B.; Rodrigues, A.C.M. Controlled crystallization and ionic conductivity of a nanostructured LiAlGePO4 glass-ceramic, J. Non. Cryst.

Solids 2009, 355, 2295–2301.

[14]

Nuernberg, R.B.; Pradel, A.; Rodrigues, A.C.M. A systematic study of glass stability, crystal structure and electrical properties of lithium ion-conducting glass-ceramics of the Li1+xCrx(GeyTi1-y)2-x(PO4)3 system, J. Power Sources 2017, 371, 167–177.


51


[15]

Page 52 of 63

Magistris, A.; Chiodelli, G.; Villa, M. Lithium borophosphate vitreous electrolytes,

J. Power Sources 1985, 14, 87–91.

[16]

Muñoz, F.; Montagne, L.; Pascual, L.; Durán, A. Composition and structure dependence of the properties of lithium borophosphate glasses showing boron anomaly, J. Non-Cryst. Solids 2009, 355, 2571–2577.

[17]

Zielniok, D.; Cramer, C.; Eckert, H. Structure/Properties Correlations in IonConducting Mixed-Network Former Glasses: Solid-State NMR Studies of the System Na2O-B2O3-P2O5. Chem. Mater. 2007, 19, 162–3170.

[18]

Kumar, S.; Rao, K.J. Lithium Ion Transport in Germanophosphate Glasses,

Solid State Ionics 2004, 170, 191–199.

[19]

Prasad, P.S.S.; Rani, A.N.D.; Radhakrishna, S. Mixed Glass Former Effect in AgI-Ag2O V2O5-P2O5 Quaternary Amorphous Solid Electrolytes, Mater. Chem.

Phys. 1990, 25, 487–499.


52

Page 53 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[20]

Raguenet, B.; Tricot, G.; Silly, G.; Ribes, M.; Pradel, A. The Mixed Glass Former Effect In Twin-Roller Quenched Lithium Borophosphate Glasses, Solid State

Ionics 2012, 208, 25–30.

[21]

Tho, T.D.; Prasada Rao, R.; Adams, S. Structure Property Correlation in Lithium Borophosphate Glasses, Eur. Phys. J. E 2012, 85, 8(1-11).

[22]

Larink, D., Eckert, H. Mixed Network Former Effect in Ion-Conducting Alkali Borophosphate Glasses: Structure/Property Correlations in the System [M2O]1/3[(B2O3)x(P2O5)1−x]2/3 (M = Li, K, Cs), J. Phys. Chem. C 2012, 116, 26162– 26176.

[23]

Moguš-Milanković, A.; Sklepić, K.; Mošner, P.; Koudelka, L.; Kalenda, P. Lithium-Ion Mobility in Quaternary Boro-Germano–Phosphate Glasses, J. Phys.

Chem. B 2016, 120, 3978–3987.

[24]

Sklepić, K.; Banhatti, R. D., Tricot, G., Mošner, P.; Koudelka, L.; MogušMilanković, A. Insights from Local Network Structures and Localized Diffusion


53


Page 54 of 63

on the Ease of Lithium Ion Transport in Two Mixed Glass-Former Systems, J.

Phys. Chem. C 2017, 121, 17641–17657.

[25]

Moguš-Milanković, A.; Sklepić, K.; Blažanović, H.; Mošner, P.; Vorokhta, M.; Koudelka, L. Influence of germanium oxide addition on the electrical properties of Li2O-B2O3-P2O5 glasses, J. Power Sources 2013, 242, 91–98.

[26]

Benett, A.E.; Griffin, R.G.; Ok, J.H.; Vega, S. Chemical shift correlation spectroscopy in rotating solids: radio frequency-driven dipolar recoupling and longitudinal exchange, J. Chem. Phys. 1992, 96, 8624-8627.

[27]

Hartmann, P.; Jana, C.; Vogel, J.; Jäger, C. 31P MAS and 2D exchange NMR of crystalline silicon phosphates, Chem. Phys. Lett. 1996, 258, 107-112.

[28]

Tricot, G.; Delevoye, L.; Palavit, G.; Montagne, L. Phase identification and quantification in a devitrified glass using homo- and hetero- nuclear solid state NMR, Chem. Comm. 2005, 42, 5289-5291.


54

Page 55 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[29]

van Wüllen, L.; Wegner, S.; Tricot, G. Structural changes above the glass transition and crystallization in aluminophosphate glasses: an in situ high temperature MAS NMR study, J. Phys. Chem. B 2007, 111, 7529-7534.

[30]

Mošner, P.; Vorokhta, M.; Koudelka, L.; Montagne, L.; Revel, B.; Sklepić, K.; Moguš-Milanković, A. Effect of Germanium Oxide on the Structure and Properties of Lithium Borophosphate Glasses, J. Non-Cryst. Solids 2013, 375, 1–6.

[31]

Henderson, G.S. The Germanate Anomaly – What Do We Know? J. Non-Cryst.

Solids 2007, 353, 1695–1704.

[32]

Evstropev, K.S.; Ivanov, O.A. Physicochemical Properties of Germanium

Glasses, in: Advances in Glass Technology, Part 2, F.R. Matson, G.E. Rindone (Eds.), Plenum Press, New York, 1963, pp.79–85.

[33]

Nelson, B.N.; Exarhos, G.J. Vibrational Spectroscopy of Cation-Site Interactions in Phosphate Glasses, J. Chem. Phys. 1979, 71, 2739–2747.


55


[34]

Page 56 of 63

Behrends, F.; Eckert, H. Mixed Netork Former Effects in Oxide Glasses: Spectroscopic Studies in the System (M2O)1/3[(Ge2O4)x(P2O5)1-x]2/3, J. Phys.

Chem. C 2014, 118, 10271-10283.

[35]

Losilla, E.R.; Cabeza, A.; Bruque, S.; Aranda, M.A.G.; Sanz, J.; Oglesias, J.E.; Alonso, J.A. Syntheses, Structures, and Thermal Expansion of Germanium Pyrophosphates, J. Sol. St. Chem. 2001, 156, 213–219.

[36]

Irvine, J.T.; Sinclair, D.C.; West, A.R. Electroceramics: characterization by impedance spectroscopy, Adv. Mater. 1990, 2(3), 132–138.

[37]

Sinclair, D.C. Characterisation of electro-materials using ac impedance spectroscopy,

Bol. SECV. 1995, 34(2), 55–65.

[38]

Hsu, C.H.; Mansfeld, F. Concerning the conversion of the constant phase element parameter Y0 into a capacitance, Corrosion 2001, 57, 747–748.

[39]

Money, B. K., & Hariharan, K. Crystallization kinetics and phase transformation in superionic lithium metaphosphate (Li2O–P2O5) glass system, Journal of Physics: Condensed Matter, 2009, 21(11), 115102.


56

Page 57 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[40]

Dabas, P., & Hariharan, K. Lithium rich phosphate glass: Crystallization kinetics, structural and conduction characteristics, Journal of Non-Crystalline Solids, 2012, 358(2), 252-260.

[41]

Money, B.K.; Hariharan, K. Lithium ion conduction in lithium metaphosphate based systems, Appl. Phys. A. 2007, 88, 647–652.


57


Page 58 of 63

TABLE CAPTION

Table 1.

Glass composition and experimental data obtained by DTA (glass transition temperature, Tg, crystallization onset temperature, Tc, and crystallization temperatures, T1 and T2, for 40Li2O-(60-x)P2O5-xGeO2 (x=0-25 mol%) series of glasses.

Table 2.

Crystalline phase composition (wt%) of studied glass-ceramics obtained by heat-treatment at different temperatures for 24 h, based upon Rietveld analysis. Note: the amount of glassy phase has not been determined and thus not been taken into account in calculating amounts of crystalline phases.

Table 3.

Proportions of 31P within the amorphous phase.

Table 4.

The fitting parameters obtained from equivalent circuit modeling of complex impedance spectra measured at 90 °C for Ge-5 glass and selected glassceramics samples prepared by heat-treatment of selected starting glasse at various temperatures for 24 h.


58

Page 59 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 5.

DC conductivity, σDC, activation energy, EDC, and pre-exponential factor, σ0*, for all investigated glass-ceramics.


59


Page 60 of 63

FIGURE CAPTION

Figure 1. DTA curves of 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glasses recorded at heating rate 10 K min-1. Glass transition temperatures, Tg, are determined from the inflection point, Tx represents the onset crystallization, T1 is defined as the temperature between Tc and T2, T2 is a maximum and Tm is melting temperature. Lines represent the temperatures of heat treatments.

Figure 2. PXRD diffraction patterns for 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glassceramics prepared by heat-treatment at various temperatures for 24 h.

Figure 3.

31P

MAS-NMR spectra of 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glass-

ceramics prepared by heat-treatment at various temperatures for 24 h, accompanied by typical spectral decomposition (dotted lines).

Figure 4. 2D 31P RFDR map (a) and 31P horizontal projections showing the presence of two different phosphate phases (b) in the crystallized glass Ge15@520.


60

Page 61 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. SEM micrographs of selected 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, glassceramics prepared by heat-treatment at various temperatures for 24 h.

Figure 6. Complex impedance spectra measured at 90 °C for (a) Ge-5 starting glass, (b) Ge-5 heat-treated at 520 °C, (c-d) Ge-10 heat-treated at 500 °C and 580 °C respectively. . Symbols (colored circles) denote experimental values, whereas solid black line corresponds to the best fit. The corresponding equivalent circuit model, comprising of multiple parallel combinations of the resistor (R) and the constant-phase element (CPE), used for fitting the data of individual sample and its interpretation is shown in each figure (defined as follows: gl-glassy phase, g-grain, gb-grain boundary, bc-beginning of crystallization and elelectrode polarization).

Figure 7. Complex impedance spectra at 90 °C for glass-ceramics prepared by heattreatment for 24 h at various temperatures: (a) Ge-15 at 590 °C, (b) Ge-20 at 570 °C, and (e-c) Ge-25 at 580 °C.


61


Page 62 of 63

Figure 8. Compositional dependence of DC conductivity, DC, at 363 K for starting glasses 40Li2O-(60-x)P2O5-xGeO2, 0x25 mol%, and glass-ceramics prepared by heat-treatment at various temperatures, T1 and T2, for 24 h. Dependence of activation energy, EDC, upon composition is presented for glass-ceramics prepared by heat-treatment at T2.


62

Page 63 of 63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TOC Graphic


63

Ionic Conductivity of Lithium Germanium Phosphate Glass

Recommend Documents