RCsumC. -L'analyse des mesures de transport sur les cristaux ioniques en termes de mod&les des di.fauts est s~mplifiee si quelques paramktres des defauts peuvent Ctre connus par d'autres experiences, comme les mesures NMR, ou par des calculs theoriques. Ce papier demontre que le calcul des parametres des dtfauts ponctuels est trks utile pour la resolution de plusieurs probltmes de transport dans les halogenures alcalins et d'argent. La croissante concurrence entre les rksultats des analyses fondires sur les mesures macroscopiques (conductiviti. ionique) et microscopiques (NMR) est montree en se refkrant aux resultats sur les cristaux qui ont la structure de la fluorite.Abstract. -The analysis of transport measurements on ionic crystals in terms of defect models is facilitated if some of the defect parameters can be determined from separate experiments, such as NMR measurements, or from theoretical calculations. This paper shows how the calculation of point defect parameters has proved useful in resolving a number of transport problems in the alkali and silver halides. The growing measure of concurrence between the results of analyses based on macroscopic (ionic conductivity) and microscopic (NMR) measurements is illustrated by reference to results for crystals with the fluorite structure. are inconsistent with those determined from diffusion measurements [7]. The reason for this is not hard to see. Alkali halide crystals generally contain an excess of divalent cation impurities like Ca2+ so that the conductivity in the extrinsic temperature range is dominated by the motion of cation vacancies. The cation migration and association parameters are therefore determined very precisely. An analysis of the whole conductivity curve using an imperfect model thus forces the parameters for the minority carriers -the anion vacancies -to assume incorrect values. If these are constrained to values consistent with diffusion measurements, then the Schottky defect formation parameters assume unrealistically high values. One of the principal difficulties about the non-linear least squares analysis of conductivity data is the high correlation that exists between some parameters. The use of more elaborate defect models inevitably means the introduction of more parameters with the result that the whole procedure tends to become rather intractable. There are two main ways out of this dilemma. One is to exploit selective doping that will accentuate one or more transport mechanisms at the expense of others, so that the precise value of the defect parameters corresponding to the latter have little effect and they may be held constant during the leastsquares analysis. The other is to make full use of other methods of determining some of the transport parameters, that is by theoretical calculation or by one of the microscopic methods of measuring defect parameters. There is one microscopic technique which provides direct information about the diffusional motion of defects and that is nuclear magnetic reso-'Article published online by EDP ...