Results and applications
Spin-exchange optical pumping
The SEOP method consists in electronic polarisation of alkali atoms (usually Rb) by optical pumping and the transfer of the polarisation to the 3He nuclei through the hyperfine interaction during collisions. Since the first demonstration by T. Chupp, this technique has been continuously developed in the USA. The use of almost unlimited power from spectrally narrowed lasers light combined with Rb-3He cells of record lifetime (≈ 300h) now leads to the routine production of NSF with 70-75% 3He polarisation in cells up to 0.5l in volume. However, the production rate is almost two orders of magnitude smaller than for the MEOP technique. With much lower alkali spin-destruction rates, K and Na could advantageously replace Rb. Because of the unavailability of convenient lasers, a hybrid approach where the Rb spin is quickly transferred to the K has been tested. For a K-rich vapour, the results show that the production rate is increased by almost an order of magnitude (about 2.4 bar.l/day).
In Europe, the SEOP programme has begun at ISIS and FZJ some years ago (ENPI - European Neutron Polarisation Initiative, HPRI-CT-1999-50016). After the ENPI initiative, the objectives of the NSF-JRA project was to complete the construction of the SEOP stations and perform trial experiments before exploring the hybrid technique and build systems to polarise 3He continuously on beams. Today, FZJ has a conventional SEOP station polarising 3He up to 62% and ISIS has replaced its 30W 2nm FWHM FAP laser with a 50W Quintessence diode in an external cavity diode laser (ECDL) providing a spectrally narrowed laser light (0.12nm FWHM). The maximum 3He polarisation has increased from 32 to 70% and beam tests have been very successful on the instruments CRISP and ROTAX. During the project, ILL has decided to work on the SEOP technique and has built an ECDL laser pumping cells up to 72% polarisation. FZJ has started working on the hybrid technique and has constructed the components for implementing spin filters on the instrument KWS-1 at FRM-II. ILL has designed and constructed a magnetostatic cavity for on-beam SEOP pumping PF1B. The measurements performed on PF1B have revealed that the 3He polarisation decreases with increasing neutron flux. This discovery will probably have a big impact in the future.
Metastability-exchange optical pumping
With the MEOP method, metastable 3He pumping takes place in a discharge at low pressure (≈ 0.7mbar) and a polarisation preserving compression by a factor of a few 1000 has to follow. The compression phase requires some technical efforts which are rewarded by the shorter spin exchange time constant and the better polarisation achieved. Before the NSF-JRA project, ILL was starting running its second-generation MEOP station (Tyrex) and providing NSF cells on instruments with almost 70% 3He polarisation. As regards HMI and ISIS, they were starting the construction of their own stations.
Together, FRM-II and ILL have constructed an optical polarimeter for measuring the absolute 3He polarisation whose design prevents the measured polarisation to be affected by spurious reflected light. The method is based upon the measurement of the amplitude of the circular polarisation of the 668nm light emitted by the plasma of a polarised gas in an optical pumping cell. The new polarimeter runs perfectly well and is routinely used at ILL.
In order to improve the performance of Tyrex, ILL has built a set of electronics and optics with which to stabilise automatically the wavelength of the light. The method consists in tuning the Yb-fibre laser units on the C8 or C9 transitions by maximising the amplitude of the fluorescence light produced by small optical pumping cells irradiated with the low-power monitoring light of each unit. ILL has also replaced the optics (polarising cube + λ/4 plate) with ones offering better performance at high power and change the shape of the electrodes placed on the optical pumping cells (OPC) for applying the discharge. With all these modifications performed, the maximum polarisation has raised from 75 to 83% in static mode and the production rate now reaches more than 15 bar.l/day when filling NSF with up to 80% polarised gas. In the meantime, FRM-II has acquired a filling station built at Mainz (called Helios) showing good performance (+70% in cells).
As regards the development of new polarisation preserving compressors, ISIS has redesigned a peristaltic compressor. The flow rate of this pump is 77 mbar.l/h at nominal speed and the maximum output pressure is 1.8 bar. The compressor has been integrated into their MEOP kit and the first tests with polarised gas have revealed a polarisation preservation coefficient of only 25%. This result is clearly not satisfactory but better results could be obtained with the use of one-way valves preventing back-flow and C-flex tubing. However, it became quite clear that the piston technique used at ILL is much more efficient and ISIS has asked ILL to build a MEOP station similar to Tyrex. As regards HMI, they have almost finished the construction of their own MEOP filling station whose performance are not expected to be comparable to those obtained at ILL.
Because the gas polarisation decays with time, a remote-type of operation is generally adopted: NSF cells are detached from the filling station and moved to the neutron beam facility. After usage, the cells are returned on the filling station and refilled with freshly polarised gas. But in some circumstances, the NSF cell cannot easily be detached from the instrument and must be replenished on-site (local-filling). This novel method has been developed and tested at ILL on the reflectometer D17. The gas of the NSF cell, housed in the evacuated detector tank, is evacuated and recovered. Then the cell is refilled through a capillary that traverses the wall of the evacuated tank via a free expansion from the buffer cell. The entire process does not last more than 30 min and the measured polarisation losses do not exceed 1%\cite. A decisive step has been accomplished when ILL has demonstrated that a filter could be maintained at a constant polarisation by repolarising small quantities of gas in pulsed mode using a special magnetic capillary. CEA-DSM and ILL now exploit this novel technique on the diffractometer D23.
Polarised 3He containers
The wall relaxation of the 3He polarisation scales with the surface-to-volume ratio and depends strongly on the quality of the inner surfaces of the containers. Aluminosilicate glasses like Supremax and Corning 1720 are known to have good relaxation times but they contain boron leading to significant transmission losses for thermal and cold neutrons. Detachable cells (MEOP) are usually made from Pyrex or quartz glass (GE224) and Cs coated. SEOP cells, on the other hand, are generally sealed, must be heated to 200-300°C and therefore require the use of a quartz that is less porous This is why our colleagues from NIST use GE180 tubes. At the beginning of the NSF-JRA project, the production of valve-sealed cells for MEOP was not reliable, with lifetimes varying between 60 and 200 hours. After many investigations at all facilities and some fruitful discussions with colleagues from the USA, we have finally adopted a more reliable recipe leading to the production of containers with long wall relaxation times (200 to 450 hours) at HMI, FRM-II, FZJ, ILL and ISIS.
HMI has investigated the influence of external magnetic fields on the relaxation behaviour of polarised 3He in NSF cells. This includes studies on the effect of the cell material, its coating, various cleaning procedures and the orientation of the cells in a guide-field as well as the search for a threshold value of the magnetic field at which the observed phenomenon starts to take place. Investigations on silica-glass cells, Cs-coated as well as uncoated, showed that even a short exposure to a strong magnetic field of 0.5 to 0.7 T results in a huge decrease of the relaxation time constant, while a careful demagnetisation procedure can restore the initial relaxation behaviour completely. It turned out that each created state proved to be stable during months. In order to better understand this effect, 2D-SQUID measurements have been performed. They confirm an enhancement of the magnetisation of the cells after an exposure to an external field. Further investigations have even shown that the relaxation drops dramatically if the cell is magnetised with a field of only 200 G. Following this discovery, Partners have taken care to apply low field for maintaining the 3He polarisation.
ILL has launched the successful construction of containers with single-crystal silicon windows minimising the diffuse scattering of neutrons and therefore the background seen in the detectors of SANS instruments and reflectometers. Because of the bounding technique used by the Hellma company for fixing the Si windows to the Pyrex bodies, these cells cannot hold large pressures (∆p ≈ 0.3 to 0.6 bar for respectively Ø6 and Ø14 cm cells) but that remains enough for SANS and reflectometry and the lifetimes are quite good (typically 240h).
ILL has also performed finite-element calculations of banana-shaped cells for the PASTIS project (XYZ polarisation technique). From these calculations, we have realised that a reinforcement of the top and bottom sides of the cells would lead to a higher pressure acceptance. Two copies have been manufactured in electronic-grade quartz-glass (GE224) by companies working with clean rooms. The first one has been pressure tested up to its destruction for checking its ability to accept pressures up to 3 bar with a security factor of 1.5. After Cs coating, the wall relaxation time of the second cell is about 400 hours.
Magnetostatic cavities
In order to render the magnetic field relaxation negligible, the magnetic field applied on the whole volume of the NSF must have a relative gradient of the order of 10-4 cm-1 or better. Neutron scattering instruments contain generally large quantities of magnetic material which is often magnetised. Moreover, many polarised neutron experiments are performed with a high magnetic field applied on the sample position and large inhomogenous stray fields must be screened. To cope with these fields, new devices have been designed for bringing cells to the instruments and for carrying out experiments successfully, even with position sensitive detectors.
From ANSYS finite-element calculations, ILL has designed, constructed and tested two transport units (called “magic boxes”) made of µ-metal and permanent magnets. They screen low environmental magnetic fields (up to 7 G), protect the users from accidental explosion of the container, do not require the transport of a battery and feature field relaxation times of 500 and 120 hours respectively for the large and small versions. The smallest box can host a standard NSF cell and only weight 6 kg.
An upgrade of this “magic box” consisting in the addition of a solenoid producing an oscillating RF field perpendicular to the static field at the 3He Larmor precession frequency has been developed at ILL and tested at both the ILL and on the instrument MIRA at FRM-II. The superposition of the fields is a circularly precessing field that can be used to return the 3He polarisation (application of the Adiabatic Fast Passage method well established in NMR). The AFP 3He flipper has been tested very successfully with a measured polarisation loss per flip less than 1.02±0.1 x 10-5. 12 copies of this box have been produced and distributed to the partners and observers. It can be noted that there is no need for a neutron spin flipper when using this box and this simplifies further the use of polarized beams.
In parallel with the preparation of the PASTIS cell mentioned above, ILL has calculated and built a set of coils able to produce a homogeneous magnetic field that can be rotated towards 3 perpendicular directions. These are XYZ pairs of coils with compensation correction coils that have been tuned using nuclear magnetic resonance free induction decay. Tests have been performed on IN3 at ILL. With the applied field rotating every 5 min toward the X, Y and Z directions, the relaxation time measured was exactly the one expected for a cell filled with 3 bar of polarised gas, i.e. ≈100 hours.
At HMI, FRM-II and FZJ, a few magnetostatic cavities have been built with the aim to start using spin filters on beam lines. HMI and TU-Aachen are finishing the construction of a detector with polarisation analysis following ILL Decpol design for the three-axis spectrometers E1 and the diffractometer Heidi. FRM-II has started the construction of specific cavities for the three-axis and neutron resonance spin-echo spectrometers and FZJ has realised a cavity (“magic box” type) to be installed on the SANS instrument KWS-1 for polarisation analysis. For instruments using high magnetic fields, a special magnetostatic cavity made from superconducting and soft-magnetic materials able to host a 3He container in front of a cryomagnet has also been built and tested very successfully at ILL.
Conclusion
Today, the world’s leading technique for polarising 3He remains MEOP in Europe but the SEOP community is making big progress and the European facilities have acquired most of the knowledge that has been developed in the USA.
The number of instruments taking advantage of the spin filters has increased very rapidly during the project. For example, the quantity of polarised gas produced for scheduled experiments has been multiplied by 3 at ILL. In most cases, the use of spin filters has increased the performance of instruments: better flux and/or resolution, less background, easier data analysis. It has also opened new possibilities, for example the investigation of magnetic nano-scale samples on powder diffractometers (D20 - ILL) and the exploitation of polarised beams at spallation sources (ISIS, SNS).
Co-ordinator: Eddy Lelièvre-Berna (ILL)
|
Supported by the EU through FP5 and FP6