The past year has seen an ever-increasing number of high pressure experiments at the ESRF: a dozen beamlines have used this parameter for many studies ranging from inelastic scattering at a few kbar to diffraction above 2 Mbar, as well as classical EXAFS with a large volume cell, the first ever EXAFS measurements on single crystals, Mössbauer spectroscopy beyond 1 Mbar, and small-angle scattering from polymers and biological samples.

In the field of diffraction, the main work has been in obtaining accurate structural information from light elements and compounds, and generally from materials in extreme conditions of pressure and temperature. Three examples taken from these fields are reported below.

Furthermore, much effort has been invested in developing a fast on-line image plate detector system for high-pressure diffraction experiments; this is reported in the section "New Techniques and Instrumentation".


Phase diagram of iron in extreme conditions of pressure and temperature

As iron is the dominant component of the Earth's core, information on its behaviour at high pressure and temperature is fundamental in the earth sciences, but despite numerous studies on this subject there is still uncertainty about its crystallographic structure at very high pressures and temperatures. The accurate determination of the phase diagram of this element has remained an experimental challenge because of the extreme conditions involved, and both shock-wave and diamond-anvil cell measurements have revealed a wealth of phase transformations at high pressure and temperatures, but there are disagreements within the geophysics community about the stability field of the various phases. In fact, recent reports are still in conflict within the 40 to 100 GPa region of the phase diagram, more particularly when considering the possible existence of a new phase in the high pressure extension of the stability field of the -phase.

Extreme conditions of static pressure and temperature in materials are obtained by combining laser heating with diamond-anvil high pressure cells. Difficulties in obtaining simultaneously uniform heating, reliable temperature measurements and very high pressures limit the sample dimensions to a few microns in thickness and a few tens of microns across. Using micro-focusing of the radiation produced by two phased undulators on beamline ID30, it has been possible to collect reliable intensity data from such micro-samples in extreme conditions of pressure and temperature. Full diffraction cones are collected with monochromatic radiation and image plates, and this using acquisition times of 10 to 15 minutes for samples of a few microns in thickness. The problems of re-crystallisation generally encountered at high temperature are largely avoided by integrating the diffraction rings collected by the area detector, and the resulting improvements in terms of resolution and data quality make possible the investigation of subtle structural details. When studying iron and various minerals, data of sufficiently high quality to perform structure refinement were collected up to 100 GPa and beyond. In the case of iron, 2D images collected at 45 GPa and 2100 K (Figure 78) revealed subtle changes from its -structure to another high pressure, high-temperature polymorph, and the quality of the data was sufficient to perform Rietveld refinement of the spectra (Figure 79). The diffracted intensities were used to refine a new orthorhombic structure, and this very important step opens the new field of "extreme conditions crystallography" where the fine determination of structural changes in an extended pressure-temperature field is now possible.


D. Andrault (a), G. Fiquet (b), M. Kunz (c), F. Visocekas (a) and D. Häusermann (c), Science, in press.

(a) Laboratoire de Physique des Géomatériaux, Institut de Physique du Globe, Paris (France)
(b) Laboratoire de Sciences de la Terre, Ecole normale Supérieure de Lyon (France)
(c) ESRF




Structural studies and recrystallisation of Ar/O2 solid mixtures at high pressure

Atomic diffusion, site ordering and recrystallisation at high pressure are important for pressure-induced phase transitions, studying the geochemistry of the inner earth and material synthesis at high pressure, but they have been scarcely investigated by experiments above 1 GPa. We have observed remarkable self-organisations in Ar/O2 solid mixtures at high pressure, which have been characterised by optical observation, Raman spectroscopy and X-ray diffraction.

At first the system forms two compounds, Ar(O2) and Ar(O2)3, and presents a site ordering when the pressure is increased. This is reminiscent of the important Au/Cu system at ambient pressure, a favourite subject of order-disorder transformation studies. Later we observed a recrystallisation from a fine-grained polycrystalline mixture of pure solid O2 and almost pure solid Ar, which led to the growth of large single-crystals of an Ar(O2)3 compound. This implies a great mobility of the atoms and molecules which is quite unexpected when vacancy mechanisms are excluded .

Solid oxygen is the only magnetic insulator among the elements, and numerous studies have investigated its phase diagram at high pressure. In particular, around 10 GPa strong changes in the properties of the solid have been related to a pairing between O2 molecules, or an intra-molecular spin pairing. From a geometric point of view, a rotating O2 molecule is very similar to an atom of argon, and they should therefore be highly miscible if there is no change in their chemical interaction. Indeed a total miscibility of O2 and Ar in the solid phase has been reported at low temperature, and we have shown that this continues up to about 8 GPa, which is in a pressure range where the oxygen bonding is changing. This work shows that the Ar/O2 phase diagram offers interesting possibilities to probe diffusion and self-organisation effects under pressure.

The Ar/O2 phase diagram at 296 K is shown in Figure 80. It was determined by studying a total of 40 different compositions in the Ar-O2 systems using optical observation and Raman spectroscopy. A detailed discussion of this diagram will be given elsewhere, so that we will presently limit our report to a summary of the X-ray measurements carried out at the ESRF. Single-crystal diffraction measurements were performed on beamline ID9 to determine the arrangement of the molecular entities in the alloys at concentrations of 50 mol% and 75 mol% O2. White beam energy-dispersive single-crystal diffraction was used to measure the structural parameters, and monochromatic angle-dispersive diffraction with image plates to visualise ordering effects. This provides a unique example of the experimental flexibility available to the ESRF users since the high-pressure cell was moved from the energy-dispersive instrument on ID9 to the angle-dispersive instrument on ID30 during the measurements to observe the recrystallisation of the sample.

For the diffraction measurements, single crystals were grown in the sample chamber. Strong reflections, including multiple higher order peaks, were obtained and they could be related by a cubic orientation matrix. The number of molecular entities in the unit cell was obtained with the use of the equation of states of Ar and O2 under the assumption that the volume should not differ much from that in a case of ideal mixing. The space-groups were inferred from observed extinctions. At 50 mol% O2, over the pressure range where only one vibron peak is observed (see Figure 80), Ar and O2 form a disordered hcp alloy. At higher pressure, where two vibron peaks are observed, the reflections match the hexagonal P63mc space-group, with Ar and O2 respectively in (2a) and (2b) sites. Hence it is reasonable to assume that the alloys, with concentration below 75 mol%, have their structure based on the hcp Ar structure, the O2 substitution being either random or on specific sites. Substituting O2 (for concentration greater than 10 mol%) for argon changes the structure from fcc to hcp, and we can now state that a site ordering occurs with pressure in ArO2 solid from a perfect substitutional disorder on a hcp lattice to a structure in which Ar and O2 occupy alternate layers. At 75 mol% O2, the space-group was identified as Pm3n with Ar and O2 respectively in (2a) and (6c) sites. The corner and body centred sites are occupied by the Ar atoms and there are two O2 molecules on each face. This is almost the -O2 structure where the molecules in the orientational disorder sites have been replaced by the argon atoms. -O2 is stable along the melting line up to 300 K, so somehow replacing rotating O2 molecules by argon atoms (which have almost identical effective pair potentials) extends the stability field of the -O2 structure.

Lastly, during the single-crystal experiment, an increase in pressure led to the formation of a fine mixture of O2 and argon so that when the cell was taken to an angle-dispersive diffraction instrument on beamline ID30, powder diffraction lines due to the two components were obtained (Figure 81). However, observation of 2-dimensional images taken as a function of time revealed the slow formation of Ar(O2)3 single-crystals at constant pressure. Since atomic ordering necessitates atomic diffusion, this observation is quite puzzling as in this case the probability of vacancies is very small. This dramatic recrystallisation sequence of Ar(O2)3 is illustrated in Figure 81.


P. Loubeyre (a), M. Jean-Louis (a), R. LeToullec (a), M. Hanfland (b) and D. Häusermann (b), submitted to Phys. Rev. Letters.

(a) Physique des Milieux Condensés, Université Pierre et Marie Curie, Paris (France)
(b) ESRF




Equations of state of 7LiH and 7LiD up to 94 GPa

The lithium hydrides have been extensively studied since the early days of solid state theory, as they are the most elementary ionic compounds, and now much attention is paid to their high pressure properties because of their important use in nuclear fusion research. Central to their high pressure behaviour is the determination of their equations of state (EOS), and we present here single-crystal X-ray diffraction measurements on 7LiH and 7LiD into the 100 GPa range performed at ID30.

These high-precision data constrain very well the equation of state of the lithium-hydrides from ambient pressure to the Mbar range, and a clear isotopic shift is observed.

The method used for our measurements is the one developed for the study of hydrogen in the 100 GPa range. We have measured the EOS of three single-crystals of 7LiD and two single-crystals of 7LiH. In the experiments below 40 GPa the sample chamber was typically 150 µm in diameter, the single crystal 50 µm long and about 20 µm thick. For those up to 100 GPa, these dimensions were 50 µm, 20 µm and about 10 µm respectively. The pressure was measured using the quasi-hydrostatic ruby scale [Mao et al.] with a precision of ± 0.03 GPa. Six classes of diffraction peaks (111, 200, 220, 311, 222 and 400) were followed in all experiments, the reflections were related by the orientation matrix of the NaCl structure up to the maximum pressure, and no effects of uniaxial strain were observed. Such single-crystal measurements provide the most accurate structural data possible and the error in the volume measurement was estimated to be less than ± 2 x 10-3 cm3/mole.

The cell parameter of the B1 structure of 7LiD versus pressure is shown in Figure 82. Three experiments were performed to cover different pressure ranges with wide overlaps, and the small dispersion in the data illustrates their high accuracy and reproducibility. The dashed line is a fit which represents the data within the experimental uncertainty, and the inset shows the enlarged low pressure part of the curve to compare our data with those published elsewhere, more particularly with the recent high pressure neutron measurements to 10 GPa [Besson et al.] where the agreement is excellent.

Similar experiments were performed on 7LiH up to 36 GPa and these show clearly an isotopic shift in the EOS which is well outside the error bars. While V/V at a given pressure decreases slowly from 1% at ambient to 0.8% at 36 GPa, P at a given volume increases from 0.3 GPa to 1 GPa over that range. As the influence of the ionic zero-point motion on the EOS is difficult to take into account in ab-initio calculations, we have used a simplified model based on the Einstein and Debye approximations for the energy of the zero-point vibration in the optical and acoustic modes, respectively. We thus estimated the difference in pressure at a given volume to be 0.5 GPa near ambient pressure and 1.2 GPa near 40 GPa.

Considering the simplicity of the model, this is in excellent agreement with our measurements, and we can therefore state that the non-negligible isotopic shift in the EOS of the lithium hydrides amounts to that expected from the zero-point vibrations. Consequently, it is important to note that this observation gives even more weight to the disturbing recent measurement of an unexpectedly small difference between the EOS of H2 and D2.

Various theoretical studies have tried to predict the behaviour of LiH at high pressure, and they show two transitions: metallisation and a B1 (NaCl) to B2 (CsCl) structural change. The present determination of the EOS does not allow a critical examination of band gap calculations, and hence of the question of metallisation, but the EOS being the derivative of the enthalpy, a good agreement between calculated EOS and the present experimental determination should give confidence in the calculated pressure for the B1-B2 phase transition.

When comparing our data with the two theoretical EOS calculations which agree best with our measurements, and reproduce quite accurately the B1-B2 phase transition when applied to the other alkali-hydrides, we find that these predict the B1-B2 transition at around 85 GPa in one case and 450 GPa in the other. Considering that one of these calculated EOS is slightly more compressible than our measured EOS, whereas the other is slightly less, we can thus confidently state that if the B1-B2 structural change does take place, it should do between these two pressures. However, since the enthalpy was not calculated for other possible structures, one cannot rule out the occurrence of other transitions within the range of diamond-anvil cells. This is an important motivation for the extension of these measurements in the 200 GPa range, or to apply new ab-initio constant pressure molecular dynamic calculations [10] to this problem.


R. Le Toullec (a), P. Loubeyre (a), M. Hanfland (b), L. Ulivi (c), F. Datchi (a) and D. Häusermann (b), to be published.

(a) Physique des Milieux Condensés, Université Pierre et Marie Curie, Paris (France)
(b) ESRF
(c) Istituto di Elettronica Quantistica, Firenze (Italy)




The structure of the -phase of molecular nitrogen

Like other diatomic molecules (O2, CO, ...) molecular nitrogen undergoes numerous structural phase transitions during varying temperature and pressure. At least five different crystalline modifications have been identified. At low temperatures and high pressures these phases are characterised by orientationally-ordered molecules, whereas at high temperatures and low pressures the molecules are orientationally disordered. At room temperature, the transformation from orientationally disordered to orientationally ordered molecules takes place at 170 kbar. The ordered phase is called e-nitrogen and its structure, determined by Mills et al. from diffraction measurements at 200 K and 120 kbar, is supposed to be rhombohedral (space group: R-3c). This is however in disagreement with Monte Carlo (MC) and molecular dynamics (MD) calculations, which favour tetragonal structures for the ordered phases.

At the ID9 beamline, we studied the structural properties of molecular nitrogen at high pressures with angle-dispersive powder diffraction (ADX) and image plates as detectors. Samples were loaded into a diamond anvil high pressure cell by condensing nitrogen gas into the gasket hole of the cell at liquid nitrogen temperatures. A diffraction image measured at 180 kbar is shown in Figure 83 and the corresponding diffraction pattern in Figure 84. Combined with the high X-ray flux of the ESRF, ADX with image plates makes it possible to measure diffraction patterns at high pressures even of weakly scattering samples like molecular nitrogen with a resolution and sensitivity never achieved before. Despite having a much larger sample, Mills et al. for example only observed the seven strongest diffraction peaks between 8 and 13°, and one peak at higher angles. Here all diffraction peaks to a maximum diffraction angle of 2-theta = 24° (limited by the opening angle of the diamond anvil cell used in the experiment) were observed. The pattern enabled us to refine the structure of the -phase. As a structural model, the rhombohedral structure proposed by Mills et al. was used. Taking into account preferred orientation along the 111-direction, a good agreement between the measured and the calculated patterns was achieved, proving that the structure of the -phase is rhombohedral and not tetragonal as suggested by the MD calculations. Structural parameters are close to the values given by Mills et al.


M. Hanfland (a), M. Lorenzen (a), C. Wassilew-Reul (a), and F. Zontone (a), Proceedings of the AIRAPT-16 conf. (Kyoto 1997), submitted.

(a) ESRF




Probing magnetism in the Mbar range

Nuclear forward scattering (NFS) of synchrotron radiation is especially suited for probing magnetism of Fe at very high pressure. Here we report on high pressure NFS studies performed at ID18 with the 14.4 keV transition of 57Fe on magnetic RFe2 Laves phases of cubic C15 structure (R = Y and Gd) as well as hexagonal C14 structure (R = Sc) at pressures up to 100 GPa (= 1 Mbar). These Laves phases are model systems for magnetism in intermetallic compounds and exhibit, as a function of the lattice parameter and the properties of R, a variation of the Fe magnetism from a localised high-moment via an itinerant low-moment to a non-magnetic behaviour. Within the pressure range of the present study, we vary the lattice parameter by more than 10%, which is more than the variation observed within the various RFe2 series, thereby explicitly probing the dependence of the Fe moment formation on the Fe-Fe distance.

High pressure was applied to the samples with diamond-anvil cells. Figure 85 shows typical spectra of YFe2 at 300 K and various pressures. The spectrum at ambient pressure exhibits a periodic beat pattern, which originates from a hyperfine field of 18.6 T at the 57Fe nuclei, which corresponds to a local Fe moment of 1.66 µB. The spectral features are slightly damped by an additional quadrupole interaction, which originates from the non-cubic local surrounding at the Fe site. With increasing pressure we observe for YFe2 (and the other investigated RFe2 systems) a decrease of the magnetic hyperfine fields (with a concomitant decrease of the Fe moment) and an increase of the quadrupole interaction as demonstrated by the spectrum at 50 GPa, where the "magnetic" beats are already strongly modified. Going further up with pressure, there is a drastic change in the spectrum at 71 GPa, now being dominated by a quadrupole interaction with a beat period about 110 ns, slightly modified by a small magnetic hyperfine field of 3 T. The spectrum at 105 GPa arises from a pure quadrupole interaction with a beat period of 100 ns. Additional studies on YFe2 at 105 GPa showed the complete absence of magnetic interactions down to temperatures as low as 15 K. From this behaviour and additional NFS spectra measured with and without external field, we can state (see Figure 86) that the magnetic ordering temperature in YFe2 is lowered from 535 K at ambient pressure to 300 K around 75GPa and to below 15 K at 105 GPa.

Similar series of NFS spectra were taken from GdFe2 with and without external fields up to 105 GPa. Gd possesses a large S = 7/2 spin moment from the 4f7 electron shell, which couples in a canted spin structure ferrimagnetically with respect to the Fe moments. The magnetic Gd sublattice enforces the magnetism in GdFe2, as reflected by the higher magnetic ordering temperature of 790 K and larger hyperfine fields at the 57Fe nucleus. The pressure variation of these hyperfine fields is shown in Figure 86 and demonstrates that GdFe2 remains magnetically ordered above 300 K up to 105 GPa. This different behaviour, in comparison to YFe2, is clearly caused by the magnetic Gd sublattice, since the structural properties and compressibilities of YFe2 and GdFe2 are very similar.

We know from additional studies that the Gd sublattice orders itself around 120 K at ambient pressure and that this ordering temperature increases strongly with pressure. From Figure 86 one may deduce that the magnetic strength of the Gd sublattice surpasses that of the Fe sublattice somewhere above 50 GPa, thereby taking over the leading role in the Fe moment formation. The different magnetic behaviour of Fe in YFe2 and GdFe2 is, in our opinion, of importance for a basic understanding of the Fe moment formation and stability in other Fe intermetallics, for instance in hard magnets like Nd2Fe14B.

In ScFe2, we observe a decrease of the magnetic ordering temperature from 500 K (0 GPa) to 300 K (51 GPa), combined with a change from a ferromagnetic (high moment) to antiferromagnetic (low moment) state. This behaviour resembles that of fcc Fe layers stabilised on Cu(Au) substrates with varying lattice parameters.

These first high pressure NFS studies of magnetism in the Mbar range yield new information about the magnetic properties of the RFe2 Laves phases and, in addition, valuable information about the local electronic properties under such extreme compression via the quadrupole interaction and the isomer shift. This wealth of information is obtained within measuring times of 10 min, 30 min and 2 h for NFS spectra at 0, 50 and 100 GPa, respectively. These qualities are, in our opinion, unprecedented by other methods. Future NFS studies are possible to even higher pressures, where for instance the stability of 4f moments may be probed.


R.Lübbers (a), H.J. Hesse (a), H.F.Grünsteudel (b), R.Rüffer (b), J.Zukrowski (c), G.Wortmann (a), to be published.

(a) FB Physik, Universität Paderborn (Germany)
(b) ESRF
(c) University of Mining and Metallurgy, Cracow (Poland)




X-ray absorption measurements on single crystals at high pressure

The primary effect of the application of pressure to a material is the reduction of the interatomic distances. This will in turn modify the intensity and the hierarchy of the interatomic forces, leading eventually to new structures. In that respect, it is a fundamental tool to discriminate between the various theoretical models and therefore to understand the properties of the chemical bonds.

In low-dimensionality solids (chain-like or layer compounds) there are two types of interactions, the intensity ratio of which is as high as one order of magnitude. Therefore, the pressure dependence of the strong and weak bonds will be very different and the intensity ratio will be very sensitive to the applied pressure. To compare theoretical calculations with experimental data, the pressure dependence of the interatomic distances (atomic positions) has to be known. Unfortunately for such low-dimensionality crystals, it is difficult or even impossible to produce powdered samples and therefore to perform powder X-ray diffraction without strong preferential orientations, making the quantitative analysis of the data difficult .

X-ray absorption spectroscopy (XAS) has been shown to be a good tool to determine the evolution of interatomic distances with pressure. Nevertheless, due to the size of the X-ray spot (500 µm) in the first generation synchrotron radiation sources, the whole experimental size (200 µm) of the high pressure cell had to be filled. This excludes the possibility of performing XAS on single crystals.

The specially designed focusing optics of beamline ID24 enables a full spot of 50 micrometers to be obtained. With such a spot, it is possible to make XAS measurements on single crystals in the diamond anvil cell (Figure 87). In this case I0 is measured inside the cell just below the sample.

We measured the EXAFS (Extended X-ray Absorption Fine Structure) spectra from the layer compound GaSe on a single crystalline sample up to 32 GPa at the Ga K edge. The size of the sample was approximately 100 x 100 x 20 µm3 and the pressure was measured in situ using the ruby fluorescence technique.

Figure 88 shows the evolution of the absorption spectrum at different pressures. Around 25 GPa a smoothing of the oscillations is observed. Up to this pressure, the Ga-Se distance decreases regularly with pressure (Figure 89). Above this pressure, a transition to a six-fold coordination occurs as evidenced by the rapid increase of the first neighbours distance.

In the low pressure phase, the experimental data has been fitted to a Murnaghan equation of state (EOS) with B0 = 110 GPa and B'0 = 5, L0 = 2.458 A (with B0 the bulk modulus, B'0 its pressure derivative and L0 the Ga-Se distance at ambient conditions). The value obtained for B0 differs significantly from the value obtained for the lattice parameter a using Brillouin spectroscopy (65 GPa). EXAFS analysis gives directly the Ga-Se distance, whereas Brillouin scattering gives a macroscopic bulk modulus. Hence, the difference between the two results may be explained by a change in the Ga-Se-Ga and Se-Ga-Ga angles.

It should be emphasised that this is the first high pressure EXAFS experiment on a single crystal. This opens a large field of measurements using the linear polarisation of the photons to study the evolution of the structural anisotropy of single crystals under high pressure.

Moreover, the size of the focused beam enables measurements at very high pressure. To perform experiments in the megabar (100 GPa) range, the hole has to be reduced down to about 50 µm or less. This remains the same order of magnitude as the size of the X-ray spot achieved during this experiment. Therefore, the megabar range is now accessible to EXAFS measurements in third generation synchrotron radiation sources, and particularly at the ID24 beamline of the ESRF.



J.P. Itié (a), A. Polian (a), M. Gauthier (a) and A. San Miguel (b), to be published.

(a) Physique des Milieux Condensés, Paris (France)
(b) ESRF