Motivations and experimental


In order to apply the RDM method to glasses, one should dispose of crystalline-based starting models presenting the same composition or at least the same formulation as the material to be modelled. During the recrystallization study inside the vitreous domain in the systems NaF/PbF2/MF3 (M = Fe, V), one phase was identified with the NaPbM2F9 composition and its structure was determined ab initio [24] from powder diffraction data in spite of the absence of suitable single crystal. The structure is built up from linear intercrossed chains of corner-linked [MF6] octahedra. It is worth noting that some Na/Pb disorder was suggested to be present in the crystalline phase. Some years later, the structure of a polytype of NaPbFe2F9 was determined from single crystal data with a tripled cell parameter c : KCaAl2F9 disclosed together with two isotypical compounds KCaFe2F9 and KCaV2F9 [25]. The KCaAl2F9 crystal structure presented some anomalies and difficulties in refinements which were suggested to be due to possible microtwinning and/or defaults in the stacking sequence with parts presenting possibly the NaPbFe2F9 structure-type. On another hand, a lot of other enneafluorides are known which present a highly flexible structure deriving from a common basis built up from isolated staircase double chains of [MF6] octahedra differently oriented (NaBaFe2F9 [26], KPbCr2F9 [27], Ba2ZnAlF9 with Zn/Al disorder [28]). With neighbouring compositions, the Ba7CuFe6F34 structure [29] belonging to the Jarlite type as well as the BaTiF5 structure [30] were also considered as possible model candidates.

Owing to this abundance of crystal structures which could serve as starting models by using the RDM method, it was decided to study glasses having exactly the NaPbM2F9 composition. Glasses were prepared by melting the anhydrous fluoride mixtures in a dry box (inert atmosphere), then the melt, in a covered platinum crucible, were cast and rolled in a bronze mould heated at 200°C. Neutron data were recorded at ILL (Grenoble) on the D4 instrument for M = Fe and V (wavelength : 0.703 Å) [31]. X-ray data were recorded on a Siemens D500 diffractometer with CuK-alpha radiation, normalized and rebuilt by interpolation as corresponding to the 0.703 Å wavelength for the modelization purposes.

The expected isomorphous replacement between Fe3+ and V3+ is well supported by the crystal chemistry in fluorides in general and in particular for the compounds listed in Table I, most of these crystal structures being used in the RDM modelling. As a rule, when a Fe3+ -based fluoride exists, the isostructural equivalent V3+ material can be prepared too with generally no more than 1% variation in cell dimension. The mean usual interatomic distances are 1.935 and 1.950 Å respectively for Fe-F and V-F in octahedra. These considerations apply exclusively to fluoride compounds because Fe3+ and V3+ cations may present a quite different behaviour in oxydes having a less pronounced ionic character than fluorides.

Table I. Cell parameters comparison between some Fe and V-based isostructural compounds.

a

b

c

alpha

beta

gamma

Ref.

FeF3 5.196 5.196 13.33 90.00 90.00 120.0

[32]

VF3 5.170 5.170 13.40 90.00 90.00 120.0

[32]

NaPbFe2F9 7.308 12.559 7.640 90.00 93.06 90.00

[24]

NaPbV2F9 7.274 12.570 7.629 90.00 92.63 90.00

this work

NaBaFe2F9 7.371 17.533 5.475 90.00 91.66 90.00

[26]

NaBaV2F9 7.372 17.555 5.491 90.00 91.60 90.00

[32]

KCaFe2F9 12.758 7.468 23.23 90.00 90.00 90.00

[25]

KCaV2F9 12.778 7.481 23.28 90.00 90.00 90.00

[25]

BaFeF5 14.919 14.919 15.218 90.00 90.00 90.00

[32]

BaVF5 14.97 14.97 15.06 90.00 90.00 90.00

[32]


-/0/+
Armel Le Bail - June 1997