- Email: [email protected]
The structure of AgICsI glass
Solid State lonics 70/7 North-Holland
SOLID STATE IOWlCS
1 ( 1994) 390-393
The structure of AgI-CsI glass H. Takahashi,
Y. Hiki
Applied Physics Group, Faculty OfEngineering, Ibaraki University, Nakanarusawa. Hitachi 316. Japan
T. Sakuma Department ofphysics, Faculty of Science, Ibaraki University, Mite 310, Japan
S. Funahashi Japan Atomic Energy Research Institute, Tokai 319-I 1. Japan
The structure of (AgI)0.55(CsI),,,,, glass has been studied by neutron diffraction. A low-Q peak suggesting the formation of intermediate range structure ( - 8 A) was observed in the structure factor S(Q). Inspection of the radial distribution functions for glass and crystal suggests that the glass can be characterized as a framework of distorted iodide tetrahedra.
1. Introduction In recent years, much attention has been focused on superionic conducting glasses. Among these, AgIcontaining glasses are frequently investigated because of their structural stability and high ionic conductivity near room temperature. AgI-AgP03 is a typical superionic conducting glass [ 11. It can be clarified that Ag and I ions are nearest neighbors in the glass, and that the structure of the glass-forming component is not changed by the addition of AgI, as confirmed by neutron [ 21 and X-ray [ 3 ] diffraction. However, the local structure of the AgI component and the conduction mechanism for the Ag ions are not known. One of the difficulties in analyzing the structure of superionic conducting glasses is that the X-ray or neutron scattering from the glassforming component is dominant compared with the scattering from the AgI component. It is known that, on the basis of the equilibrium phase diagram [ 41, the AgI-CsI system has two crystalline phases: CsAgJ, and CszAgIJ. On the other hand, glass formation in AgI-based halide systems has been reported [ 5 1, and in this system the glass 0167-2738/94/$07.00
compound is formed without glass-forming oxides such as BzOJ or P205. The glass-forming range in this system is 50-60 mol% AgI. Since the AgI-CsI glass has no complex glass-forming component, this type of glass is considered to be useful for investigating structure of the AgI component in the glass, especially in the short- and middle-range structure regime. The purpose of the present investigation is to measure the structure factor of AgI-CsI glass by the neutron diffraction, and to characterize the structure of AgI-containing glasses.
2. Experimental
procedure
Reagent-grade anhydrous AgI and CsI were used as the starting materials. After referring to the phase diagram [ 41, appropriate amounts of AgI and Csl were mixed and melted in a Pyrex crucible at 400’ C. When the melt became transparent, the sample was poured on a stainless-steel block and pressed further by another block. By this method, rapid quench and glass formation of a thin sample can be realized. The glass sample so prepared was transparent and col-
0 1994 Elsevier Science B.V. All rights reserved.
391
H. Takahashi et al. / The structure ofAgl-Cd glass
orless. However, the sample gradually became opaque when exposed to moisture in the air. It was confirmed that the sample became crystalline as its transparency was lost. Therefore samples were stored in a dry-box. The plate-shaped sample was coarsely ground and sealed into a cylindrical aluminium container for the neutron diffraction experiment. Silica gel was further placed at the top and bottom of the container to prevent the crystallization on contact with humidity. The inner diameter and length of the container were 14 mm and 50 mm, respectively. To diminish the scattering from the container, the thickness of the container wall near the sample position was reduced. The upper and the lower ends of the container were shielded by cadmium. The neutron diffraction measurements were carried out at room temperature with the double-axis diffractometer installed at JRR-2 at the Japan Atomic Energy Research Institute (JAERI ). The reflection from the (002) plane of a pyrolytic graphite crystal monochromator was used to produce a beam with a wavelength A= 1.0 A. In a typical experiment, the neutron intensity was recorded in the range 4” <219x 110” at 0.1’ intervals. The neutron absorption coefficient of the sample was estimated experimentally by measuring the transmittance through the sample at 2&O”. 3. Experimental
results
Diffraction measurements were made for: (A) the glass sample immediately after preparation, (B) the sample preserved in an aluminium container for one month, (C) the sample after heating to 100” C (the glass transition temperature of the glass determined by thermal analysis: about SO’C) and (D) the empty aluminium sample container. Fig. 1 shows the observed diffraction profiles for (A)-(C) above. Diffraction peaks marked * represent those from the aluminium container. Extra peaks can be observed in profile (B ). Crystallization of the glass sample must have occurred, because of the moisture remaining in the container and the relatively low glass transition temperature. These peaks were enhanced by heating the sample (profile (C) ). Two crystalline phases, CszAgIs and CsAg,13, can be formed in the AgI-CsI system. The extra peaks ap-
n 4”
0
20
40 60 2 8 ldeg)
El0
100
Fig. 1. The neutron diffraction profile for ( AgI)O.SS( CSI)~.,~ under several experimental conditions (see text).
pearing in profiles (B) and (C) seem to represent the superposition of the Bragg peaks from these compounds. On the contrary, profile (A) contains only peaks from the aluminium container. The crystalline precipitate is negligible immediately after glass formation. The profile for (AgI)o.,,(CsI)o.,, glass was obtained by subtracting the aluminium profile from profile (A). Experimental scattering profiles were smoothed by the spline fitting method, and several experimental corrections were made as described in previous work [ 61. The observed structure factor S(Q) for the glass is shown in fig. 2; note the small peak observed around Q=O.8 A-‘. This corresponds to a structure unit of the order of 8 A. Such a low-Q peak is also observed in the structure factor of the superionic conducting glass AgI-AgPO, [ 21. On the basis of the RMS simulation [ 7 1, it is clarified that the origin of the low-Q peak in AgI-AgP03 glass is the density fluctuation of the PO network. Although the true origin of this peak is not clear in this glass, some type of density fluctuation might occur.
4. Discussion The pair distribution
function
g(r)
can be ob-
392
H. Takahashl et al. / The structure ofA@-Cd
Fig. 2. Structure
factor of (A~I),,,,(CSI)~.,,
glass.
-
2
4
6
8
10
r Ii1 Fig. 3. Pair distribution
function
of (AgI)0.55(C~I)0.45 glass.
tained from the Fourier transform
of S( Q) as follows:
g(~)=1+[1/(2~*~p0)1 X
I
[S(Q)-llQsin(Qr)
dQ,
(1)
where p. is the number density in the glass. The number density of the present glass was not measured, so the average value for crystalline CszAg13 and CsAg,I, is adopted. In fig. 3, g(r) determined from the observed S(Q) by using eq. ( 1) is shown. The crystal structures of CszAgI, and CsAgJ, were reported by Brink et al. [ 8,9]. In order to compare with the AgI-CsI glass structure, the superposed radial distribution functions of Cs2Ag13 and CsAg213
glass
crystals are also shown in the upper part of the figure. From the radial distribution functions of the crystals, the interionic distance between Ag and I is seen to be distributed around 2.8 A. The distributions around 4.0 8, and 4.5 A mainly represent the Cs-I and I-I correlations, respectively. Interionic distances between cations (Ag-Ag, Ag-Cs and CsCs) are distributed around 4.5 A and beyond 5.5 A. The pair distribution function g(r) for the glass can be interpreted on the basis of interionic distances in the crystal. From the neutron scattering cross-sections and the ion concentrations, the contributions of Ag-I, I-I, Cs-I, Ag-Cs, Ag-Ag and CsCs pairs to g(r) is 0.31, 0.22, 0.20, 0.13, 0.10 and 0.04, respectively. The summation of partial pair distributions of Ag-I, I-I and Cs-I thus contributes the main part of g(r). The first peak in g(r) at 2.8 A is assigned to Ag-I correlation. The second peak consists of three parts: a central peak at 4.3 8, and shoulders at 3.9 and 4.9 A. The shoulder at 3.9 A can correspond to Cs-I correlation. On the other hand, the I-I correlation at 4.5 8, in the radial distribution for the crystals is split into two parts (at 4.3 8, and 4.9 A) for the glass. As mentioned earlier, the AgAg, Ag-Cs and Cs-Cs contributions to the pair distribution function are small. The second peak of g( r) mainly consists therefore of Cs-I and I-I correlations. A Gaussian curve fit to the radial distribution function of the glass was made to clarify the relation between ion paris and interionic distances (fig. 4). The full line indicates the experimental radial distribution function, broken lines show each of the Gaussian curves and the superposed curve. The radial distribution function below 6 8, is presented by eight Gaussian curves. The central positions of the Gaussian curves and their assignments to ion pairs are summarized in table 1. On the whole, the interionic distances of Ag-I and Cs-I pairs scarcely vary from crystal to glass. In the crystal, the interionic distances between I ions are nearly the same (4.5 A). Thus, I ions form regular tetrahedra with Ag ions at their centers. On the other hand, the interionic I-I distance is distributed around 4.3 and 5.0 8, in the glass. From the viewpoint of the interionic distances of Ag-I and I-I, it can be concluded that a distorted tetrahedron of AgI, ions is formed. Cs ions occupy the positions between the distorted AgL, tetrahedra at the Cs-I distance of 3.9
H. Takahashi et al. / The structure ofAgI-Csl
-
12
a 6
L
4
2
3
393
ence of Cs ions neighboring AgI, tetrahedra. So the low-Q peak in the S(Q) is assigned to the density
10
L 52
glass
4
5
6
7
fluctuation of the AgI framework. On the basis of the pair distribution function obtained, consideration of the detailed structure of this glass is now in progress. Superionically conducting a-AgI is a typical crystal containing distorted iodide tetrahedron. The ionic conductivity of AgI-CsI glass is of the order of lop3 S/cm, while that of the crystalline AgI-CsI system is less than lo-’ S/cm. It is interesting to investigate whether the formation of aAgI-like structure in the glass has any effect on its ionic conductivity. It is thus important to study dynamical properties of this glass to understand the effect of structure on ionic conduction.
I & Fig. 4. Observed and fitted curves for the radial distribution function of (AgI),.,,( CSI)~,~~glass. Table 1 Central positions for the Gaussian curves and their assignments to ion pairs. No.
Central position (A)
Assigned ion pair
1
2.85 3.84 4.32 4.61 4.98 5.45 5.79 6.20
1st Ag-I 1St cs-I 1st I-I 1st Ag-Cs, 1st Ag-Ag 1St I-I 2nd Ag-I, 1st Cs-Cs 2nd Cs-I, 1st Cs-Cs
2 3 4 5 6 7 8
A. Since the number of I ions is twice the number of Ag ions, isolated AgI, tetrahedra would not exist. Rather the distorted iodide tetrahedra are considered to be connected by face-sharing to form the framework structure in the glass. The density fluctuations of AgI framework occur because of the pres-
Acknowledgements The authors thank Dr.N. Minakawa (JAERI) for his help with the neutron scattering measurements.
References [ 1] J.P. Malugani, A. Wasniewski, M. Doreau, G. Robert and R. Mercier, Mater. Res. Bull. 13 ( 1978) 1009.
[ 21 M. Tachez, R. Mercier, J.P. Malugani and A.J. Dianoux, Solid State Ionics 20 (1986) 93. [ 3 ] A. Musinu, G. Piccaluga and G. Pinna, J. Non-Cryst. Solids 106 (1988) 70. [ 4 ] J.N. Bradley and P.D. Greene, Trans. Faraday Sot. 63 ( 1967 ) 424. [ 51 K. Kadono, A. Yasuyoshi, K Nakano, K. Kinugawa and H. Tanaka, J. Ceram. Sot. Japan 100 (1992) 233. [ 61 H. Takahashi, S. Takeda, S. Harada and S. Tamaki, J. Phys. Sot. Japan 57 (1988) 562. [ 71 L. Borjesson, R.L. McGreevy and W.S. Howells, Philos. Mag. B 65 (1992) 261. [ 81 C. Brink and C.H. MacGillavry, Acta Crystallogr. 2 ( 1949) 158. [9] C. Brink, N.F. Binnendijk and J. van de Linde, Acta Crystallogr. 7 ( 1954) 176.
SOLID STATE IOWlCS
1 ( 1994) 390-393
The structure of AgI-CsI glass H. Takahashi,
Y. Hiki
Applied Physics Group, Faculty OfEngineering, Ibaraki University, Nakanarusawa. Hitachi 316. Japan
T. Sakuma Department ofphysics, Faculty of Science, Ibaraki University, Mite 310, Japan
S. Funahashi Japan Atomic Energy Research Institute, Tokai 319-I 1. Japan
The structure of (AgI)0.55(CsI),,,,, glass has been studied by neutron diffraction. A low-Q peak suggesting the formation of intermediate range structure ( - 8 A) was observed in the structure factor S(Q). Inspection of the radial distribution functions for glass and crystal suggests that the glass can be characterized as a framework of distorted iodide tetrahedra.
1. Introduction In recent years, much attention has been focused on superionic conducting glasses. Among these, AgIcontaining glasses are frequently investigated because of their structural stability and high ionic conductivity near room temperature. AgI-AgP03 is a typical superionic conducting glass [ 11. It can be clarified that Ag and I ions are nearest neighbors in the glass, and that the structure of the glass-forming component is not changed by the addition of AgI, as confirmed by neutron [ 21 and X-ray [ 3 ] diffraction. However, the local structure of the AgI component and the conduction mechanism for the Ag ions are not known. One of the difficulties in analyzing the structure of superionic conducting glasses is that the X-ray or neutron scattering from the glassforming component is dominant compared with the scattering from the AgI component. It is known that, on the basis of the equilibrium phase diagram [ 41, the AgI-CsI system has two crystalline phases: CsAgJ, and CszAgIJ. On the other hand, glass formation in AgI-based halide systems has been reported [ 5 1, and in this system the glass 0167-2738/94/$07.00
compound is formed without glass-forming oxides such as BzOJ or P205. The glass-forming range in this system is 50-60 mol% AgI. Since the AgI-CsI glass has no complex glass-forming component, this type of glass is considered to be useful for investigating structure of the AgI component in the glass, especially in the short- and middle-range structure regime. The purpose of the present investigation is to measure the structure factor of AgI-CsI glass by the neutron diffraction, and to characterize the structure of AgI-containing glasses.
2. Experimental
procedure
Reagent-grade anhydrous AgI and CsI were used as the starting materials. After referring to the phase diagram [ 41, appropriate amounts of AgI and Csl were mixed and melted in a Pyrex crucible at 400’ C. When the melt became transparent, the sample was poured on a stainless-steel block and pressed further by another block. By this method, rapid quench and glass formation of a thin sample can be realized. The glass sample so prepared was transparent and col-
0 1994 Elsevier Science B.V. All rights reserved.
391
H. Takahashi et al. / The structure ofAgl-Cd glass
orless. However, the sample gradually became opaque when exposed to moisture in the air. It was confirmed that the sample became crystalline as its transparency was lost. Therefore samples were stored in a dry-box. The plate-shaped sample was coarsely ground and sealed into a cylindrical aluminium container for the neutron diffraction experiment. Silica gel was further placed at the top and bottom of the container to prevent the crystallization on contact with humidity. The inner diameter and length of the container were 14 mm and 50 mm, respectively. To diminish the scattering from the container, the thickness of the container wall near the sample position was reduced. The upper and the lower ends of the container were shielded by cadmium. The neutron diffraction measurements were carried out at room temperature with the double-axis diffractometer installed at JRR-2 at the Japan Atomic Energy Research Institute (JAERI ). The reflection from the (002) plane of a pyrolytic graphite crystal monochromator was used to produce a beam with a wavelength A= 1.0 A. In a typical experiment, the neutron intensity was recorded in the range 4” <219x 110” at 0.1’ intervals. The neutron absorption coefficient of the sample was estimated experimentally by measuring the transmittance through the sample at 2&O”. 3. Experimental
results
Diffraction measurements were made for: (A) the glass sample immediately after preparation, (B) the sample preserved in an aluminium container for one month, (C) the sample after heating to 100” C (the glass transition temperature of the glass determined by thermal analysis: about SO’C) and (D) the empty aluminium sample container. Fig. 1 shows the observed diffraction profiles for (A)-(C) above. Diffraction peaks marked * represent those from the aluminium container. Extra peaks can be observed in profile (B ). Crystallization of the glass sample must have occurred, because of the moisture remaining in the container and the relatively low glass transition temperature. These peaks were enhanced by heating the sample (profile (C) ). Two crystalline phases, CszAgIs and CsAg,13, can be formed in the AgI-CsI system. The extra peaks ap-
n 4”
0
20
40 60 2 8 ldeg)
El0
100
Fig. 1. The neutron diffraction profile for ( AgI)O.SS( CSI)~.,~ under several experimental conditions (see text).
pearing in profiles (B) and (C) seem to represent the superposition of the Bragg peaks from these compounds. On the contrary, profile (A) contains only peaks from the aluminium container. The crystalline precipitate is negligible immediately after glass formation. The profile for (AgI)o.,,(CsI)o.,, glass was obtained by subtracting the aluminium profile from profile (A). Experimental scattering profiles were smoothed by the spline fitting method, and several experimental corrections were made as described in previous work [ 61. The observed structure factor S(Q) for the glass is shown in fig. 2; note the small peak observed around Q=O.8 A-‘. This corresponds to a structure unit of the order of 8 A. Such a low-Q peak is also observed in the structure factor of the superionic conducting glass AgI-AgPO, [ 21. On the basis of the RMS simulation [ 7 1, it is clarified that the origin of the low-Q peak in AgI-AgP03 glass is the density fluctuation of the PO network. Although the true origin of this peak is not clear in this glass, some type of density fluctuation might occur.
4. Discussion The pair distribution
function
g(r)
can be ob-
392
H. Takahashl et al. / The structure ofA@-Cd
Fig. 2. Structure
factor of (A~I),,,,(CSI)~.,,
glass.
-
2
4
6
8
10
r Ii1 Fig. 3. Pair distribution
function
of (AgI)0.55(C~I)0.45 glass.
tained from the Fourier transform
of S( Q) as follows:
g(~)=1+[1/(2~*~p0)1 X
I
[S(Q)-llQsin(Qr)
dQ,
(1)
where p. is the number density in the glass. The number density of the present glass was not measured, so the average value for crystalline CszAg13 and CsAg,I, is adopted. In fig. 3, g(r) determined from the observed S(Q) by using eq. ( 1) is shown. The crystal structures of CszAgI, and CsAgJ, were reported by Brink et al. [ 8,9]. In order to compare with the AgI-CsI glass structure, the superposed radial distribution functions of Cs2Ag13 and CsAg213
glass
crystals are also shown in the upper part of the figure. From the radial distribution functions of the crystals, the interionic distance between Ag and I is seen to be distributed around 2.8 A. The distributions around 4.0 8, and 4.5 A mainly represent the Cs-I and I-I correlations, respectively. Interionic distances between cations (Ag-Ag, Ag-Cs and CsCs) are distributed around 4.5 A and beyond 5.5 A. The pair distribution function g(r) for the glass can be interpreted on the basis of interionic distances in the crystal. From the neutron scattering cross-sections and the ion concentrations, the contributions of Ag-I, I-I, Cs-I, Ag-Cs, Ag-Ag and CsCs pairs to g(r) is 0.31, 0.22, 0.20, 0.13, 0.10 and 0.04, respectively. The summation of partial pair distributions of Ag-I, I-I and Cs-I thus contributes the main part of g(r). The first peak in g(r) at 2.8 A is assigned to Ag-I correlation. The second peak consists of three parts: a central peak at 4.3 8, and shoulders at 3.9 and 4.9 A. The shoulder at 3.9 A can correspond to Cs-I correlation. On the other hand, the I-I correlation at 4.5 8, in the radial distribution for the crystals is split into two parts (at 4.3 8, and 4.9 A) for the glass. As mentioned earlier, the AgAg, Ag-Cs and Cs-Cs contributions to the pair distribution function are small. The second peak of g( r) mainly consists therefore of Cs-I and I-I correlations. A Gaussian curve fit to the radial distribution function of the glass was made to clarify the relation between ion paris and interionic distances (fig. 4). The full line indicates the experimental radial distribution function, broken lines show each of the Gaussian curves and the superposed curve. The radial distribution function below 6 8, is presented by eight Gaussian curves. The central positions of the Gaussian curves and their assignments to ion pairs are summarized in table 1. On the whole, the interionic distances of Ag-I and Cs-I pairs scarcely vary from crystal to glass. In the crystal, the interionic distances between I ions are nearly the same (4.5 A). Thus, I ions form regular tetrahedra with Ag ions at their centers. On the other hand, the interionic I-I distance is distributed around 4.3 and 5.0 8, in the glass. From the viewpoint of the interionic distances of Ag-I and I-I, it can be concluded that a distorted tetrahedron of AgI, ions is formed. Cs ions occupy the positions between the distorted AgL, tetrahedra at the Cs-I distance of 3.9
H. Takahashi et al. / The structure ofAgI-Csl
-
12
a 6
L
4
2
3
393
ence of Cs ions neighboring AgI, tetrahedra. So the low-Q peak in the S(Q) is assigned to the density
10
L 52
glass
4
5
6
7
fluctuation of the AgI framework. On the basis of the pair distribution function obtained, consideration of the detailed structure of this glass is now in progress. Superionically conducting a-AgI is a typical crystal containing distorted iodide tetrahedron. The ionic conductivity of AgI-CsI glass is of the order of lop3 S/cm, while that of the crystalline AgI-CsI system is less than lo-’ S/cm. It is interesting to investigate whether the formation of aAgI-like structure in the glass has any effect on its ionic conductivity. It is thus important to study dynamical properties of this glass to understand the effect of structure on ionic conduction.
I & Fig. 4. Observed and fitted curves for the radial distribution function of (AgI),.,,( CSI)~,~~glass. Table 1 Central positions for the Gaussian curves and their assignments to ion pairs. No.
Central position (A)
Assigned ion pair
1
2.85 3.84 4.32 4.61 4.98 5.45 5.79 6.20
1st Ag-I 1St cs-I 1st I-I 1st Ag-Cs, 1st Ag-Ag 1St I-I 2nd Ag-I, 1st Cs-Cs 2nd Cs-I, 1st Cs-Cs
2 3 4 5 6 7 8
A. Since the number of I ions is twice the number of Ag ions, isolated AgI, tetrahedra would not exist. Rather the distorted iodide tetrahedra are considered to be connected by face-sharing to form the framework structure in the glass. The density fluctuations of AgI framework occur because of the pres-
Acknowledgements The authors thank Dr.N. Minakawa (JAERI) for his help with the neutron scattering measurements.
References [ 1] J.P. Malugani, A. Wasniewski, M. Doreau, G. Robert and R. Mercier, Mater. Res. Bull. 13 ( 1978) 1009.
[ 21 M. Tachez, R. Mercier, J.P. Malugani and A.J. Dianoux, Solid State Ionics 20 (1986) 93. [ 3 ] A. Musinu, G. Piccaluga and G. Pinna, J. Non-Cryst. Solids 106 (1988) 70. [ 4 ] J.N. Bradley and P.D. Greene, Trans. Faraday Sot. 63 ( 1967 ) 424. [ 51 K. Kadono, A. Yasuyoshi, K Nakano, K. Kinugawa and H. Tanaka, J. Ceram. Sot. Japan 100 (1992) 233. [ 61 H. Takahashi, S. Takeda, S. Harada and S. Tamaki, J. Phys. Sot. Japan 57 (1988) 562. [ 71 L. Borjesson, R.L. McGreevy and W.S. Howells, Philos. Mag. B 65 (1992) 261. [ 81 C. Brink and C.H. MacGillavry, Acta Crystallogr. 2 ( 1949) 158. [9] C. Brink, N.F. Binnendijk and J. van de Linde, Acta Crystallogr. 7 ( 1954) 176.