Analysis of the local order in the glacial state of triphenyl phosphite by neutron diffraction A. Hedoux, a J. Dore, b Y. Guinet, a M. C. Bellissent-Funel, c D. Prevost, a M. Descamps a and D. Grandjean b a Laboratoire de Dynamique et Structure des Mate ´riaux Mole ´culaires, UPRESA 8024 CNRS, UFR de Physique, Ba ˆ timent P5, Universite ´ des Sciences et Technologies de Lille, 59655, Villeneuve d’Ascq Ce ´dex, France b Physics Laboratory, University of Kent, Canterbury, UK CT2 7NR c Laboratoire Le ´on Brillouin, UMR 12 CNRS-CEA, CEA/Saclay, 91191, Gif-sur-Yvette Ce ´dex, France Received 21st June 2002, Accepted 25th September 2002 First published as an Advance Article on the web 15th October 2002 A neutron diffraction study over a wide Q-range (Q max ¼ 16 A ˚ 1 ) has been made in order to analyze the change of the local order of triphenyl phosphite through the isothermal transformation between the supercooled liquid into the glacial state at different temperatures T a . Direct evidence is given from the analysis of the time- dependence of diffraction pattern, both in the low Q- and high Q-regimes that the structural changes observed during the glaciation are the signature of a crystallization process. Consequently, the transformation of thesupercooled liquid into the glacial state can be interpreted as an abortive crystallization into a sub-microcrystalline state. Introduction First-order structural phase transitions are common in crystal- line solids whereas first-order liquid–liquid phase transitions are excessively rare in pure compounds. The existence of liquid polymorphs is beginning to be recognized and a growing inter- est is devoted to the first-order polyamorphic transitions. These phase transitions are principally observed by applying a high pressure, 1,2 and are generally associated with a change of density. The evidence of a low-temperature phase in triphe- nyl phosphite (TPP) at atmospheric pressure, 3,4 distinct from the glass, the supercooled liquid and the normal liquid was considered as an original manifestation of polyamorphism, and a real opportunity to analyze such a situation. This so- called glacial phase, was obtained via a first-order transition from the supercooled liquid state, 3–10 either by slow heating from the glass or by maintaining the temperature isothermally above the calorimetric glass transition (T g ¼ 201.8 K 6 ) in the range 210 K–235 K. The opportunity to form the intriguing glacial phase of TPP by a mere variation of temperature has given rise to numerous experimental investigations 3–19 leading to controversial descriptions of the glacial state. Various experimental results (Raman spectroscopy, 9,13,15,16 X-ray dif- fraction, 12 inelastic neutron scattering 15 and differential scan- ning calorimetry 17 ) converge into a description of the glacial state in terms of micro- or nanocrystallized domains of the stable crystalline phase, mixed with non-transformed super- cooled liquid. From this description, the origin and the relative stability of this state were explained from a rapid nucleation rate leading to a heavily nucleated state which frustrates further crystallization. 17 Evidence for rapid nucleation associa- ted to the glaciation process was reported in other works 18,19 and then supports the interpretation of the glaciation as an abortive crystallization inherent to a high density of nuclei. In this context the glacial state was described as a two-phase (amorphous/nanocrystalline) system predominantly composed of non-transformed supercooled liquid if it was prepared by isothermal ageing in the 210–222 K range. Not so far from this description, Kivelson et al. have suggested that the glacial state was apparently amorphous, 3–5 and have inter- preted this phase in terms of a ‘‘ defect-ordered phase ’’. The origin of such a picture of the glacial state, emanating from the ‘‘ frustration-limited theory ’’ (FLD) thermodynamic the- ory of supercooled liquid, 20,21 is fundamentally different from the frustration by a high density of nuclei. The formation of a ‘‘ defect-ordered phase ’’ would be connected to the inability of the system to tile the whole space by periodically replicating a locally preferred structure, i.e. to a structural frustration. Such a locally preferred local structure minimizes some local free energy in a given region of the pressure–temperature phase diagram and is associated with a close-packing arrangement of molecules. In this context the local structure in the stable crystalline phase could be different from the locally preferred structure. This latter should be characterized by an icosa- hedral-like symmetry as it can be expected from the example of close-packing of hard spheres, 22 and then the glacial state should be denser than the liquid and crystalline states as suggested by Kivelson et al. 5 from experimental investigations. Consequently the supercooled liquid and the glacial states in TPP were associated respectively with the low-density amor- phous phase and the high-density amorphous phase of H 2 O. 5 More generally, the viscous slowdown observed near T g in fragile glassformers could be a consequence of packing frustration. The picture of structural frustration is the support for the interpretation of small-angle scattering investigations 14 that have revealed structural organization on a mesoscopic scale (80 A ˚ ) which would have no relationship with the crystalline structure. Demirjian et al. 19 have described the glacial state, from nuclear magnetic resonance (NMR) experiments inter- preted in the frame of the FLD theory, as a plastic crystal composed of domains of nanocrystals. However no direct 5644 Phys. Chem. Chem. Phys., 2002, 4, 5644–5648 DOI: 10.1039/b206019a This journal is # The Owner Societies 2002 PCCP Published on 15 October 2002. Downloaded by Univ Lille 1 on 08/12/2015 07:46:23. View Article Online / Journal Homepage / Table of Contents for this issue