Large dynamic range small-angle diffractometer

D22 Polarisation

Since June 2006 D22 has been equipped with a supermirror transmission polariser and RF spin flipper to provide a polarised incident beam. More recent developments to implement spin-analysis of the scattered beam using polarised 3He are in progress. This document details the specification of the polarisation components and provides a guide to the setup of polarisation and analysis experiments on D22. This document describes how to set up D22 and gives a short reminder of important commands.

1. Polarised incident beam

The polariser consists of a 1.2 m long, FeSi m=3.6 supermirror coating on 0.5 mm single crystal silicon substrate. The 1.2 m long mirror is in reality composed of 5 different mirror pieces, precision cut and butted up to each other to make the final length. The mirror is housed in a closed magnetic circuit magnetised by a line of NdFeB permanent magnets which provide a surface field of ~0.4 T and ~0.2 T at the position of the mirror centre, half way between the two rows of magnets. The mirror housing and magnetic circuit are shown schematically in figure 1. Photographs of the polariser during assembly and in place in the casemate are shown in figure 2. The mirror angle was determined by a compromise between available space, beam cross section, mirror quality, and desired polarisation performance, and is defined by the machining of the magnet cavity at 2° to the nominal beam direction. The magnet cavity is lined with B4C to absorb the reflected and unwanted spin state. The polariser is not installed in a guide cavity as at some other institutes and therefore flux is lost due to this "hole" in the neutron guide. The seriousness of this depends on the wavelength and collimation instrument setting used but a 1.2 m gap in the guide should have a transmission of around 0.9. The polariser covers a beam area of 40x40 mm compared to the guide dimensions of (40x55 mm) i.e. 0.72 of the available guide area. This means that for a 100 % efficient polariser (i.e. all wrong spin states are reflected and absorbed and all correct spin states are transmitted), the polariser maximum transmission will be approximately:

Tmax↑ = 0.5 · 0.9 · 0.72 = 0.32 

of the available non-polarised neutron flux.

Figure 1

Schematic representation and magnetic field circuit of the polariser mirror housing.

Figure 2

1.2m long polariser magnet assembly, mirror and polariser / guide section installed in the casemate on a common translation motorised table.

1.2. Characteristics of the transmission polariser

Our initial calculations (incorrectly) assumed a perfect reflectivity curve for the m = 3.6 supermirror, i.e. R↓ = R↑ = 1 up to m = 0.6 (Silicon), R↓ = 1, R↑ = 0 up to m = 3.6, R↓ = R↑ = 0, m > 3.6. Using the relation:
λ = θc / 0.1 m  
this suggests that (zero divergence) polarised neutrons should be available from 5.5 Å until both spin states are reflected by the silicon substrate at 40 Å. Figure 3a shows the measured flipping ratio and polarisation with the mirror at 2° and for a long collimation (17.6 m), i.e. very small beam divergence ~0.1°. Clearly, beam polarisation is only of high quality for wavelengths longer than ~10 Å where the measured flipping ratio exceeds FR > 10. Since beam the beam divergence is small, the main cause of the poor polarisation at shorter wavelengths is the poor reflectivity of the mirrors at large m-values. In reality, the mirrors used here have a reflectivity which falls from R=1 @ m = 0.5 to the sharp cutoff at R = 0.45 @ m = 3.6.

Figure 3

Flipping ratio and beam polarisation for the D22 polariser installed at an angle of 2°, as designed, and at 1.5° as of beginning 2007.

As an intermediate measure to try to improve beam polarisation at shorter wavelengths we have tilted the polariser housing by 0.5°, reducing the mirror angle to 1.5°. Figure 3b shows that this significantly improves the beam polarisation at shorter wavelengths but cuts down the overall beam cross-section and therefore flux by a further 0.75. New supermirrors with better reflectivity to higher m-values are in production. Beam polarisation is heavily dependent on the incident angle of the neutron on the polarising mirror. This means that the over all beam polarisation depends on the subsequent collimation where divergence can vary from ~0.10° @ Col = 17.6 m to ~1.2° @ Col = 1.4 m. A further effect to consider is possible depolarisation by reflection from any Nickel guides used after the polariser. We do not have a full data set to describe both these phenomena but some representative data are shown in figure 4. The flipping ratio decreases at shorter collimation lengths as suspected. Local contacts / users should characterise the polarisation quality of the instrument for all the configurations desired prior to commencing their experiment.

Figure 4

Flipping ratio vs. collimation length for 10 Å neutrons and 2° mirror angle. Shorter collimation lengths result in worse beam polarisation due to increased beam divergence at the polariser and depolarisation due to the magnetic Nickel guides.

In MAD, use the command files "polin.cmd" by typing "start polin" at the MAD command line. This single line command file positions the PGTRANS motor stage underneath the polariser / guide housing to a value ~ -199 mm.
To remove the polariser and put the guide back start the command file "polout.cmd" by typing "start polout" at the MAD command line which sets the motor PGTRANS to a value of 0 mm.

All polarisation components on D22 are arranged with "blue-up" such that the Earth?s field always contributes to the guide field and helps eliminate problems of weak guide field regions and even allows us to get away with regions of no additional guide field, for example, though complex sample environments or the short secondary flight path between sample and 3He filter. The main guild field along the collimation section is made from c-shaped pieces of rolled mild steel covering just over half the circumference around the rotating collimation barrel. This means the guide field remains constant while either collimator or guide sections can be rotated to make the desired collimation length. The top and bottom edges of the c-shaped magnetic field return path are lined with ferrite magnets. This magnetic circuit gives a field of approximately 10 Gauss at the beam position, sufficient to preserve the neutron polarisation. To provide guide field up to the installed sample environment we provide "clip-on" guide field sections to cover the instrument nose, as shown in Figure 5. These should be mounted as required for each particular experiment.

Figure 5

Additional "clip-on" guide fields for the instrument "nose".

An RF spin flipper is used to invert the neutron spin direction with respect to the guide field. This has several advantages over flippers such as Mezei flippers as no material is present in the beam and can operate over a wide wavelength range (e.g. 4 ? 40 Å). In principle, once the RF flipper has been "tuned" by adjustment of "crocodile" guide field and RF field amplitude then no more adjustments should be necessary. In practice, however, since the flipper is close (~60 cm) to the sample position tuning may be necessary depending on the (magnetic) sample environment installed and strength of any additional guide fields.

Figure 6

RF spin flipper in the nose section of the instrument, close to the sample position.

The crocodile field coils should (already) be connected to the controllable power supply PS3 via the pair B connectors next to the sample area. PS3 should be setup so as to give 3 A to the crocodile field coils. The RF flipper control box should (already) be connected to the controllable power supply PS2 via connectors A. PS2 should provide 18 V to the RF flipper control box. The command file "rf_flipper_setup" run at the MAD command line automatically sets up the crocodile field and RF amplitude power supplies. A red light indicating power should show on the RF flipper control box.

The RF can be turned on/off either manually on the front panel of the control box or remotely using a 0 to 5 V signal to the remote input. We usually connect the digital-analogue converter on the MAD control computer to provide the switching voltage using the command "dac 0" (off) or "dac 5" (on). For ease, command files have been written called "flipper_on" and "flipper_off". These can be started from the Mad command line and from within other experiment command files.

A relatively new development on D22 is the provision for 3He spin analysis. With the neutron optics group we have bought one of the largest 3He cells available for neutron scattering experiments consisting of a cylindrical quartz tube of lengths 10 cm or 15 cm with 12 cm diameter single crystal Silicon faces, as shown in figure 7a. A large area cell is necessary, and should be positioned close to the sample so as to cover as large a solid angle as possible. At the same time the cell should be positioned as far as possible from any magnetic sample environment so that stray magnetic fields do not to depolarise the cell. The existing instrument configuration and spaces available dictate that the 3He cell in its large "magic-box" homogeneous field cavity should be positioned up against the large gate-valve within the main detector vacuum tank. We have made a prototype installation and tests of a 3He cell installed in the detector tank and mounted on a translation table so that it can be moved in and out of the scattered beam. Cable connections for the magic-box power supply and integrated RF3He flipper are passed though vacuum feed-through connectors, as shown in figure 7b. This is apparatus must be removed from the detector tank for non-3He analysis experiments. Care must also be taken to ensure the detector does not advance closer than ~4 m from the sample to avoid collision with the table. There is currently NO hardware safety stop installed to prevent this.

Figure 7

3He cell with 12cm Ø single crystal Silicon windows. The 3He cell sits inside the "magic box", all under vacuum inside the D22 detector tank. The magic-box assembly can be translated in and out of the scattered beam.

We have performed several tests and have begin to run scheduled user experiments using this setup. Figure 8 shows a radiographic image of the polarised 3He cell on the D22 detector. Here a stack of pyrolytic graphite was placed at the sample position to over illuminate the cell by its massive diffuse small angle scattering of the incident unpolarised beam. Figure 8 shows the transmission though the cell to be approximately 1/2 that of the fly-past beam around the sides of the cell. Interestingly, the radiographic image even shows attenuation contrast due to the quartz cylindrical cell walls, capillary fill valve and NMR test coil. The image shown in Figure 9 underlines restrictions as to the positioning of the detector for polarisation-analysis SANS experiments. The still relatively small size of the cell and its positioning ~1 m from the sample means that the detector cannot be placed closer than ~10 m before shadowing effects from the edge of the cell occur. This, along with the restriction of using longer wavelengths for better beam polarisation restricts polarisation-analysis experiments on D22 to the lower-q end of the SANS range (λ >10Å, det > 10 m).

Figure 8

Radiographic image of the 3He cell illuminated by diffuse scattering from pyrolytic graphite at the sample position.

We currently need approximately 4 h for installation of the 3He cell into the detector tank and approximately 2 h to remove it. It requires at least two people to lift the table and magic box into position. Local contacts and users must foresee significant technical support from both the instrument technician and the 3He group.

Once the table and magic box have been lifted into position connections to the magic-box field, RF 3He flipper and translation motor should be tested. The guide field in the magic box should be set to 10.5 Gauss (~1.8 A), or as otherwise advised by the 3He group. This ensures the correct parameters for the 3He RF flipper supply - again, take advice from the 3He group. The 3He group will deliver a polarised cell, pressurised as specified by the local contactor users. The gas pressure determines the cell opacity, Q, and is a product of the cell thickness x pressure x wavelength. For example, figure 9 shows that if we require the 3He to be ~99 % efficient at analyzing then we need an opacity of ~40. This can be provided by a cell pressurised to 0.4 bar · 10cm cell thickness · 10 Å neutron wavelength = 40. Polarisation, transmission and the P2T figure of merit are shown as a function of opacity, Q, in figure 9.

Figure 9

Calculated polarisation, transmission and P2T figure of merit for polarised 3He cell. Opacity is the product of cell pressure · cell thickness · wavelength.