High pressures

Christian Vettier started at ILL as a PhD student in 1972, working on antiferromagnetic materials at high pressures and low temperatures; he used the CNRS's facilities. His project involved experiments at ILL, so it was therefore only natural that the CNRS produce pressure cells for the ILL.

The constraints were as follows. Helium was the only fluid capable of transmitting hydrostatic pressure, transparent to neutrons and acceptable at high pressures and low temperatures. Inelastic neutron scattering measurements required relatively high-volume samples (2-3 cm³), and this limited the pressure available to 5-6 kbar (0.5-0.6 GPa).

Jean Paureau, an engineer at CNRS/SNCI, was working on the design of pressure cells in aluminium alloy, and on their use conditions [1]. A CNRS technician (G. Dampne) was one of the team.

These cells were known familiarly as "bombs" - they had a tendency to explode when exposed to high pressure. Raoul Mathieu (ILL), C. Vettier and J. Paureau therefore performed a series of safety tests before allowing the devices to be used on the ILL instruments. Christian Vettier remembers:

The cell to be tested was surrounded with shielding and placed in a steel Dewar flask filled with liquid nitrogen. The pressure in the cell was increased to rupture point. As the explosion occurred, the whole set-up blasted off (the helium gas being expulsed to the floor) in a cloud worthy of Cape Canaveral.

High-pressure work with neutrons became a reality and has since gone on to major success.

In 1973 Rudolf Mossbauer created a "Sample Environment" service including high pressures. Raymond Serve headed the service and explains:

Originally we bought in the high pressure cells and materials. The Americans were best in quality at the time but I didn't know any English; I was better in German. I therefore went for Nova Suisse - their cells were held to be "less efficient" but came with technical information in French or German. Nova Suisse had been through the 14-18 war and used technology of the period, but their prices were reasonable!" [Sic !]

It seems that Serve's relations with the scientists were not always easy; although nobody contested his technical skills, he was often critical of their aims. He had André Valenti in his team as draughtsman, and then Louis Mélési as a technician from 1981.

In 1989 Klaus Gobrecht took over the pressure cells group. The Annual report of that year sings the praises of his predecessor, Christian Vettier.

According to Christian Vettier, the first high pressure experiments ILL were performed at ambient temperatures on the D10 neutron diffractometer. For low temperatures you had to adapt the ILL Orange cryostats, given the size of the cells and to enable the pressure to be changed at low temperatures (see the 1979 Annual report). That's how a first Ø100 cryostat was produced, with a sample well measuring 100 mm in diameter instead of 49 or 70.

According to C. Vettier, the only competition at the time was from an Austrian group in contact with Ruepp Lechner, but they were working on more complicated type of cell. In 1976, the British (Colin Carlile) signed a contract with the CNRS/SNCI for the supply of several cells.

From then on pressure measurements with helium and low temperatures were almost routine at ILL. Louis Mélési (1982), who had had been posted temporarily to ILL to work on cryostat modifications, was eventually taken on full-time to deal with the workload involved (preparing, maintaining and operating the cells).

As Pierre Andant explains:

Raymond Serve was busy with high pressures (there was sufficient demand) as well as high temperatures. He took on a technician on a temporary basis (Louis Mélési) and the plans for the high pressure cells were produced by the Reactor Division design office (Antoine Valenti).

As an example of one of the experiments which could only be performed at the ILL at that time, we can cite the project to measure the structure of an organic compound (TTF-TCNQ) at 0.5 GPa (5 kbar) on the D8 single-crystal neutron diffractometer [2]. The helium cell was compact enough to be installed at the centre of the Euler cradle on the instrument, but it had to be able to turn in different directions. This was made possible using a compressor connected with a very long and relatively flexible metal tube [3]. Alain Filhol hasn't forgotten the details:

Christian had set up a pump unit with loads of tubes and valves... he started it all up and gave me a few quick - inevitably succinct - instructions. He headed off to another instrument, saying: "Make no mistake - it could explode!" - which was really reassuring! Well, in the end it didn't explode, and the experiment worked fine.

The 5-6 kbar provided by the gas cells were far from sufficient for the scientists. This led to another series of pressure cells being developed in parallel by J. Paureau, in collaboration with D.B. McWhan (Bell Labs), who came to spend a sabbatical year at the CNRS in 1974-1975 [4]. These were 'clamp' cells, with the pressure being generated by a liquid rather than a gas. Clamp cells can reach higher pressures with a far lower level of risk, given that far less energy is stored in the cell (liquids, unlike gases, are incompressible).

By using aluminium oxide (Al₂O₃) for the body of the cell, 30 kbar (3 GPa) can (in principle) be attained, at the price however of a smaller volume sample(0.5 cm³). The real pressure inside the cell is estimated by measuring the lattice parameters of a sample of NaCl placed inside the cell below or above the sample being investigated. In the early days (1978-1980) the cell was brought up to pressure at the CNRS. Later the ILL acquired a 6-7 kbar compressor unit and cells (Louis Mélési). The first neutron experiment was carried out by B. Renker in 1980.

This was innovative work again by the ILL/CNRS team, as there was no other expert group interested at that time in developing this very difficult technique.
This type of cell is limited by the relatively small sample volume it offers and the very narrow 'window' (the part that is transparent to neutrons partie la plus transparente aux neutrons). Another type of clamp cell (30 kbar - 3 GPa) was therefore developed (by Louis Mélési, André Valenti, and Christian Vettier) requiring a cryostat with a larger internal diameter (‘deep throat’). The cell started to be used after 1987, as the Annual report for that year says that the 30 kbar cells were still not operational. Christian Vettier explains:

Those Al₂O₃ were complex, and difficult to use. They started to deform after a time and the CNRS had not made plans for renewing them. Jean Paureau left at the end of 1979 and the programme came to a halt, as ILL didn't have the resources to keep it going.
Two Al2O3 clamp cells still exist at ILL in 2015, of the McWhan type (Bell Labs, 2 and 3 GPa). They have rarely been used since the arrival of Paris-Edinburgh cells.

(L. Mélési, A. Valenti, C. Vettier, R Serve)

The ILL thus had gas cells which were easy to use but limited in pressure (5 kbar, or 0.5 GPa), and more powerful clamp cells (20-30 kbar, or 2-3 GPa) presenting too many constraints (high background, small sample volume and therefore too weak a signal).

It was therefore decided to produce ‘small’ clamps providing an intermediate range of pressure (10 to 15 kbar at atmospheric pressure, i.e. 1 to 1.5 GPa). The new clamps used Fluorinert coolant to transmit the pressure. Fluorinert is an inert liquid used in clean rooms; it becomes vitreous at low temperatures, remaining hydrostatic in behaviour.

Several different materials have been used in their production, including marging steel, "null matrix Ti-Zr" or copper-beryllium alloys, always with great success. These alloys share much the same properties, but TiZr was invented to avoid producing Bragg peaks in the neutron detectors (zero coherent scattering cross section). When the latter cannot be used, i.e. above about 100°C, hardened CuBe (2% Be with heat treatment) is used instead.

[Click here to see an experiment being set up]

Al₂O₃ clamps are capable of 20-30 kbar (2-3 Gpa) but at the expense a very narrow angular opening for the beam window, and this was really not sufficient for some types of research. A solution would be a design of diamond anvil cells already used with X-rays but, for neutrons, the diamonds would have to be enormous, thus both hard to find and with a prohibitive cost.
The issue was addressed in two ways:

- Werner Kuhs needed a cell for high-pressure crystal-structure determination thus both compact enough to accomodate the eulerian cradle of a four-circle neutron diffractometer and with a very wide angular opening. Together with H. Ashbars, he successfully adapted a tried-and-tested X-ray diamond anvil design, replacing the diamond with sapphire (Al₂O₃).

• they 1st reached the 25 kbar mark (2.5 GPa) in 1990 [5] and later 29 kbar (2.9 GPa), for single crystal diffraction [5];
• the cell was equipped with a Peltier cooling thus allowing for variable temperature studies;
• later on a cryostat especially designed for that pressure cell which was successfully used on the 4-circle diffractometer D9 at temperatures down to 120 K (Werner Kuhs and John Archer). The sample could also be slightly heated up [6];
• finally, a more compact design of the sapphire anvil cell made it possible to place it inside an orange cryostat. This substantially increased the accessible temperature range [7].

- Iin 2004, S. Klotz, G. Hamel and J. Frelat unveiled a new 200-tonne compact hydraulic press, designed for both neutron and X-ray experiments,  the Paris-Edinburgh press [8]. The pressure available varied from 10 to 25 GPa, depending on the volume of the sample. The ILL bought several of these cells and the number of high pressure experiments doubled between 2005 and 2015.

This type of cells being quite bulky they cannot fit inside a standard Orange Cryostat. Thus, in 2005, the ILL developed a cryostat specifically for them and capable of reaching 4K in less than 8 hours. They are too large for the eulerian cradle of today's neutron diffractometers.

Under hydrostatic pressure (identical pressure in all directions) a crystal will deform progressively, but it will only break if there is a change of structure (ice cubes, for example, will break coming out of the fridge because of a change in the structure of the ice).

Uniaxial stress involves compressing a solid between two surfaces, allowing the solid to deform in the other directions. This not only changes the distances between the atoms but also the symmetry of the material (if you squash a cube you get a parallelepiped with a square base). This is why materials are less resistant to uniaxial stress than to hydrostatic pressure; applying a few kilobars to a sample without breaking it is something of a challenge. Pure metals, for example, deform irreparably beyond a few hundred bars.

A lot of experiments use uniaxial stress to change the symmetry of the sample, to favour, for example, one or another type of magnetic or crystalline domain. A uniaxial pressure cell was therefore developed in 1976 in collaboration with the CNRS (Ange Draperi and Jean Paureau). Its development was continued later by Louis Mélési [9].

Christian Vettier remembers:

The mechanism applying the uniaxial stress was simple and precise (we measured the force applied, knowing the dimensions of the contact surface of the sample); it could easily be adjusted or reversed, even with the sample maintained at low temperature in the cryostat. The user could therefore 'call up' the level of compression required: “500 bars, if you please, Mr Vettier!” There was one memorable moment with Bill Stirling on IN2 High pressures[10,11]. We were trying to lower the hexagonal symmetry of a praseodymium crystal, by applying a slight force on the hexagonal base of the crystal; lowering the symmetry should result in the appearance of magnetic order which does not exist below a certain level of deformation of the hexagon.
 So there was Bill Stirling calling the compression value, Keith McEwen watching the ratemeter, and me, perched at the top of a ladder just above the cryostat on instrument IN2, applying the force required with a torque wrench. Well, despite the acrobatics, it all went very well indeed!

In 2014 users asked ILL for a rod applying a weak uniaxial force (less than 120 N / 12 kg), in order to check the twinning in their superconducting crystals and better understand the changes in the electronic phases. E. Bourgeat-Lami, S. Turc and E. Lelièvre-Berna therefore developed a sample-stick which allows the force to be changed at any temperature between 1.5 and 300 K using a micrometer gauge. The anvils are in sapphire, to reduce the background as much as possible.

The ILL and the CNRS in Grenoble have not worked together on high pressures since 1990. As we saw above, the ILL bought a series of Paris-Edinburgh anvil cells between 2005 and 2010. Since then ILL has again started investing in high pressures as users come up with new requirements, particularly for work with biological systems and the magnetic phases of skyrmions.

Biological systems don't need to use high pressures (< 7 kbar) but they do need separators to avoid polluting the sample, as well as special geometries to optimise the scattering technique being used (small angles, for example, or back-scattering or time-of-flight). These projects are carried out in collaboration with other experts and European neutron centres.