Non-Arrhenius conductivity in glass: Mobility and conductivity saturation effects.

Extreme non-Arrhenius dependence of the ionic conductivity in optimized fast ion conducting glasses has been observed. When all the chemical factors controlling the ionic conductivity in glass have been optimized, the conductivity fails to reach the values expected, >0.1 (Ωcm)−1 at 298 K. A new series of glasses zAgI+(1−z) [0.525Ag2S+0.475B2S3:SiS2] have been measured for the first time and are found to exhibit a non-Arrhenius conductivity, the extent of which increases the greater the AgI content. Such behavior is believed to be a new feature of optimized fast ion conducting glasses and will be a critical obstacle to overcome if the conductivity of such systems is to ever reach the values needed for optimum device performance. Disciplines Ceramic Materials | Materials Science and Engineering Comments This article is from Physical Review Letters 76 (1996): 70–73, doi:10.1103/PhysRevLett.76.70. Posted with permission. This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/mse_pubs/67 VOLUME 76, NUMBER 1 P H Y S I C A L R E V I E W L E T T E R S 1 JANUARY 1996 Non-Arrhenius Conductivity in Glass: Mobility and Conductivity Saturation Effects Joseph Kincs* and Steve W. Martin Department of Materials Science & Engineering, Iowa State University, Ames, Iowa 50011 (Received 19 June 1995) Extreme non-Arrhenius dependence of the ionic conductivity in optimized fast ion conducting glasses has been observed. When all the chemical factors controlling the ionic conductivity in glass have been optimized, the conductivity fails to reach the values expected, .0.1 sV cmd21 at 298 K. A new series of glasses zAgI 1 s1 2 zd f0.525Ag2S 1 0.475B2S3:SiS2g have been measured for the first time and are found to exhibit a non-Arrhenius conductivity, the extent of which increases the greater the AgI content. Such behavior is believed to be a new feature of optimized fast ion conducting glasses and will be a critical obstacle to overcome if the conductivity of such systems is to ever reach the values needed for optimum device performance. PACS numbers: 66.30.Hs, 66.30.Dn Fast ion conduction (FIC) in glass has been studied for some time, and much effort has been directed at obtaining high conductivity in glass. Recent success in sulfideand silver-doped glasses has pushed the maximum room temperature conductivity in glass up to 1022 sV cmd21 [1]. Concomitant with this success has been the clarification of the structural and dynamic models used to understand FIC in these “superionic” glasses. Our work, for example, has clearly identified the wide composition dependence of the ionic conductivity with both structural and conduction energetics features of these glasses [2]. Other work has shown the intimate interplay between composition, structure, and the dynamics of the FIC in these glasses [3]. In all of this work, the question still remains of how high the ionic conductivity can be pushed in these glasses. For example, does the limit of 1022 sV cmd21 represent a fundamental limit that will not be overcome, or do the calculations that have been made earlier [4], where a conductivity of 1 to 100 sV cmd21 at room temperature is predicted, still hold promise that more glass chemistry optimization must be done before the limit is reached? In this Letter, we show that by using all the available knowledge that links ionic conductivity to glass chemistry and structure a new feature in the composition and temperature dependence of ionic conductivity in glass arises that may well limit the maximum conductivity that is obtainable in glass. When all the features of the glass chemistry and composition have been carefully optimized to obtain the highest conductivity in glass, the conductivity exhibits a strong non-Arrhenius temperature dependence that reduces the conductivity at room temperature some 1 to 2 orders of magnitude below that predicted from low temperature (subambient) conductivities. We believe this behavior to be an as yet undiscovered ubiquitous behavior of all superionic FIC glasses and points to another feature of ionic conduction in glass that must be fully understood in order to make any more progress in optimizing the conductivity in these glasses. Indeed, this behavior may well point to a fundamental device limitation for these glasses. The purpose of this paper is to report new measurements on a series of new silver iodide-doped silver thioborosilicate glasses that were specifically designed to yield optimum ionic conduction in glass. Glasses of general composition zAgI 1 s1 2 zd 3 fxAg2S 1 s1 2 xdB2S3:SiS2g were prepared by batch melting AgI with previously prepared xAg2S 1 s1 2 xdB2S3:SiS2 glasses in vitreous carbon crucibles in a high quality O2and H2O-free glove box at ,850 ±C and quenching into 1 2 mm 3 25 mm disks in stainless steel molds held near the Tg of the glass, ,350 ±C. The xAg2S 1 s1 2 xdB2S3:SiS2 glasses were prepared from reagent grade Ag2S and SiS2 (99.9%, Cerac, Inc.) and B2S3 prepared in this laboratory [5]. Conductivity measurements were made using a high quality impedance spectroscopy facility over the frequency range of 0.1 Hz to 32 MHz and from 100 to 600 K [4]. Complex plane analysis was used to determine the dc conductivity of these glasses. Tg’s were determined using a PE-DSC 4 at 20 ±C/min. It has been widely shown that, due to their high electronic polarizability, silver cations always exhibit conductivities in glass some 1 to 4 orders of magnitude higher than any of the alkali ions [1]. Similarly, sulfide glasses, first discovered by Levasseur et al. [6], show conductivities some 3 to as many as 10 orders of magnitude higher than any corresponding oxide glass. More recently, the effect of doping FIC glasses with halide salts, especially AgI, can increase the conductivity some 2 to 3 orders of magnitude [1]. Finally, it has also been shown that mixing different glass formers such as SiO2 and B2O3 produces nonlinear increases in the conductivity for reasons that are not completely understood, and this has been termed the mixed-glass former effect [1]. Using these observations, it follows that high ionic conductivity in glass, if not the highest yet reported, should be 70 0031-9007y96y76(1)y70(4)$06.00 © 1995 The American Physical Society VOLUME 76, NUMBER 1 P H Y S I C A L R E V I E W L E T T E R S 1 JANUARY 1996 FIG. 1. Glass forming region for ternary Ag2S 1 SiS2 1 B2S3 glasses. Glasses were quenched to room temperature in a stainless steel mold. found among glass compositions chosen in the series AgI 1 Ag2S 1 SiS2 1 B2S3. Although wide compositions of glass formation were not found, Fig. 1 shows that glasses in the ternary Ag2S 1 SiS2 1 B2S3 could be prepared. It was observed that at the SiS2:B2S3 ratio of 1:1, and a Ag2S fraction of 60 mole %, the strongest glass former was observed. This glass was then used as a host for the AgI doping and as in many other AgI-doped glasses [1], 40 mole % of AgI could be doped into the glass before devitrification was observed. The glasses reported in this paper therefore belong to the compositional series yAgI 1 s1 2 zd fxAg2S 1 s1 2 xdB2S3:SiS2g, where 0 # z # 0.4 and x ­ 0.525. Other glasses were prepared and studied and will be reported on separately. The present series is the highest conducting and most strongly glass forming. Figure 2 shows the Arrhenius plots of the conductivity for these glasses along with a few other glasses in both this family and others to show the level of conductivity increase that the current series exhibits. Figure 2 shows that these glasses do indeed exhibit the highest yet reported of all conductivities in glass at room temperature, a result quite surprising in itself, except that the glass chemistry was specifically designed to yield this result. Table I shows that the conductivity at room temperature reaches ,4 3 1022 sV cmd21 for z ­ 0.4 and is combined with a Tg of 501 K s65 Kd. The Tg’s reported in Table I are the highest ever reported for a AgI-doped FIC glass and even though Tg decreases with z, they remain exceptionally high. These two features of high conductivity and Tg are combined with the property that these glasses are exceptionally stable in both air and water. Even though they comprise some 50 at. % of SiS2 and B2S3, both of which are exceptionally chemically unstable, the resulting FIG. 2. Arrhenius plots of the ionic conductivity for glasses studied in this work and compared to those for other Li and Na conducting glasses. Notice that for the poorer conducting glasses, the Arrhenius plots have a straight slope, whereas the optimized Ag conducting FIC glasses have significant curvature at highest temperatures. glasses are very chemically durable. These three features make these glasses particularly attractive for device fabrication and use. Most dramatic about these glasses, however, is the fact that Fig. 2 shows that their conductivities are exceptionally non-Arrhenius. A dashed line on the z ­ 0.4 glass data shows that the room temperature conductivity is some 2–3 orders of magnitude less than that predicted. This behavior has been reported before for other low Tg “oxysalt” FIC glasses, where the non-Arrhenius behavior was associated with the dynamic temperature dependent restructuring of the I anion “sublattice” [7]. Such restructuring was proposed to be associated with the low Tg’s of these glasses, ,100 ±C. By annealing the glasses and presumably densifying the glass to the point TABLE I. Conductivity parameters for optimized zAgI 1 s1 2 zd f0.525Ag2S 1 0.475B2S3:SiS2g glasses. sdcs298 Kd Low temperature (extrapolated sdcs298 Kd activation energy from low T ) (actual) zAgI Tg(K) (eV) fsV cmd21g fsV cmd21g 0 593 0.33 0.0014 0.0010 0.1 576 0.32 0.0020 0.002 0.2 548 0.31 0.0071 0.003 0.3 525 0.28 0.0116 0.004 0.4 501 0.25 0.0406 0.006