Free volume and dissociation eects in fast ion conducting glasses J. Swenson a, * , A. Matic a , C. Karlsson a , L. B orjesson a , W.S. Howells b a Department of Applied Physics, Chalmers University of Technology, S-412 96 G oteborg, Sweden b Rutherford±Appleton Laboratory, Chilton, Didcot, OX11 0QX, UK Abstract We have investigated the structural changes that occur on CsI and CsI±AgI doping in AgPO 3 using neutron dif- fraction and Raman scattering experiments. To obtain further insights into the microscopic structure of the glasses we have applied the reverse Monte Carlo (RMC) method to produce structural models that are in quantitative agreement with the neutron diraction data, as well as with the experimental density and applied physical constraints. The aim was to examine the structural role of the immobile (compared to Ag ) Cs and I ions for the increased ionic conductivity upon CsI doping. It is clear from the RMC and Raman results that the Cs ions coordinate oxygens more than Ag , and in this way Cs contributes to a partial dissociation of the Ag ions from the non-bridging oxygens. The iodine ions have a similar role by forming largely covalent bonds to silver ions, leading to a further decrease of the Ag±O coor- dination. The introduction of the salt ions expands the host glass network. We suggest that both the electrostatic binding energy, E b , and the elastic strain energy, E s , associated with the ionic motion decrease with increasing CsI content due to the partial dissociation of the silver ions from the non-bridging oxygens and the expansion of interstices which are larger migration pathways for the silver ions. Ó 2000 Elsevier Science B.V. All rights reserved. PACS: 66.10.Ed; 61.43.Fs; 61.12.-q 1. Introduction Understanding the microscopic mechanism for ion conduction in glasses is a longstanding prob- lem in glass science [1]. Several properties of ion conducting glasses still lack satisfying explana- tions and none of the models that have been proposed in the literature are, so far, able to ac- count for all the experimentally determined structural and dynamical ®ndings, to our knowl- edge. One of the unsolved topics concerns the ef- fect a mixing of cations may have on the ionic conductivity. It is, for example, well known that a mixing of mobile cations reduces (up to ®ve orders of magnitude) ionic conduction, at least for non- salt doped glasses (the mixed mobile ion eect) [1,2], although the phenomenon is still not fully understood, to our knowledge. Another property may occur when mobile cations are mixed with immobile cations [3,4]. It has, for example, been observed that the silver ion conductivity in glassy AgPO 3 increases by four orders of magnitude on adding 19 mol% of the non-conducting salt PbI 2 [3,4]. A similar conductivity increase (although less) occurs for CsI doping [5]. Thus, it is evident Journal of Non-Crystalline Solids 263&264 (2000) 73±81 www.elsevier.com/locate/jnoncrysol * Corresponding author. Tel.: +46-31 772 5680; fax: +46-31 772 2090. E-mail address: f5xjs@fy.chalmers.se (J. Swenson). 0022-3093/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 3 0 9 3 ( 9 9 ) 0 0 6 7 0 - 5