WAVE – An innovative magnet devoted to spintronics

Abstract

The LLB is part of a large project aiming at the development of experimental tools available for the spintronics community. This includes the design and construction of vector magnets for neutron and X-ray scattering (deployed on the Léon Brillouin-Orphée and Synchrotron SOLEIL TGIRs: neutron diffractometer 6T2 and XMRS Sextants). For neutron scattering, a very innovative design has been developed, relying solely on the use of vertical axis coils. This magnet called WAVE (for Wide Aperture VEctor) is now available at the LLB-Orphée for the user community.

Keywords

WAVE magnet spintronics diffraction LLB vector 3D transport measurements magnetism superconductors

1. Introduction

Recently, the LLB has been involved in a large project called WAVENEXT [5], aiming at the development of experimental tools that will be available to the spintronics community, by designing, developing and building vector magnets (systems capable of applying a magnetic field in any direction) adapted to neutron and X-ray scattering. This project is co-developed with another CEA institute, the IRFU/DACM, the SOLEIL synchrotron, the Néel Institute (Grenoble) and a French industrial partner, SigmaPhi.

These magnets, as well as the complementarity of the neutron and X-ray probes, will make it possible to further study magnetism and in particular the anisotropy of magnetic materials in the form of thin films. The main constraint in the design of these magnets is to keep a large portion totally free of the scattering plane, to allow the incident and scattered beams of neutrons and/or photons to travel over a wide angular range (up to backscattering). Such system is currently not commercially available.

This project has received several fundings, from C Nano Île-de-France, ANR, and Labex NanoSaclay. For neutron scattering, a very innovative design has been developed, relying solely on the use of vertical axis coils [3]. This magnet called WAVE (see Fig. 1) is now available at the LLB for the user community. For X-ray diffraction, a hybrid prototype is currently being studied [5].

Fig. 1.

Schematic view of the WAVE magnet. Its dimensions are 2.2 m height (including power lines support) and 770 mm in diameter at the widest point (588 mm at the beam axis).

2. Magnet design

The originality of the design lies in the use of looped stray fields between several coils in order to create a very homogeneous and orientable field vector in all axes at the sample point. The magnetic design of WAVE (see Fig. 2) is based on an innovative concept developed at the DACM CEA Institute. It consists of 16 NbTi solenoids (see Table 1 for their characteristics), all with a vertical axis, designed by Pr. Guy Aubert (patent FR12 62 070, US extension 14/105,711) [1]. They are nested and glued in an aluminum “cold mass”, indirectly cooled by liquid helium circulating in a thermosiphon loop activated by a cold head [6,7]. This original arrangement allows for a very wide aperture for the neutron beam (220° horizontally, ±10° vertically), which is crucial for neutron scattering experiments. WAVE is optimized for a 10 mm spherical sample in the equatorial plane. All the coils are made with very precise continuous windings. A rectangular NbTi superconducting wire of 1.08 mm × 0.68 mm from Oxford Superconducting Technology is used here and delivers a current density of up to $J = 250$ A/mm².

Fig. 2.

Artistic view of the magnetic design.

2.1. Horizontal field component

The horizontal field component $(B_{x}, B_{y})$ is created by the stray fields of 12 coils, in green color on the Fig. 2, arranged regularly around the sample chamber well, on either side of the equatorial plane. These 12 coils are divided into 3 groups of 4 coils (one of them is highlighted in green in Fig. 3). Each group has two coils above and two coils below the diffraction plane in an antisymmetric manner (the beam axis represented by a red arrow in Fig. 3). The 4 coils of the same group are mounted in series but in opposite way to each other so that the stray fields are grouped together (see Fig. 4). This creates a uniform field in the center of the group, which is also the sample position (see Fig. 5). Each group is identical and positioned one next to the other around the sample axis. A power supply is dedicated to each group of 4 coils.

Fig. 3.

Sketch of the 12 green coils that realize the horizontal field component. Black rectangle is the cutting plane used for the Fig. 5.

The use of 3 groups of 4 coils (while 2 groups would have been sufficient) improves the homogeneity of the field [2] at the sample position (less than 0.1% on a spherical sample of 1 cm³). The input and output connections are arranged on the opposite sides of the equatorial plane of WAVE (see Fig. 4). At 0.25 T, the variation of the measured horizontal field is 0.2 gauss on a 5 mm radius spherical sample.

Fig. 4.

Electrical connections and field direction for each coil controlling the horizontal component of the field at the sample position.

Fig. 5.

Sketch of the magnetic field lines visible in a vertical cutting plane. Abscissa and ordinates are distances in mm. The black point shows the sample position.

Fig. 6.

One of the 12 “green” coils used to create the horizontal component.

2.2. Vertical field component

The vertical field component (Bz) is produced by four flat coils, two Helmholtz coils (in orange in Figs 2 and 7) and two active shield coils (in blue color in Figs 2 and 7). At 0.25 T, the typical measured variation of the vertical field component is 0.5 gauss on a 5 mm sphere radius.

Fig. 7.

Electrical connections and field direction for each coil controlling the vertical component of the field vector at the sample position.

Fig. 8.

Sketch of the magnetic field lines visible in a vertical cutting plane. Abscissa and ordinates are distances in mm. The black point shows the sample position.

Fig. 9.

Left: one of the “orange” coils visible in Fig. 7 used to create the vertical component. Right: one of the “blue” coils visible in Fig. 7 used to create an active shielding. We can see the reserve of wire kept to be able to connect it, inside the coil.

WAVE’s coils are powered by 4 current sources (provided by SigmaPhi Electronics) via high-Tc superconducting conductors. The current leads used are 4 HTS110 250A and 1 HTS110 1000A for the return. Their position is optimized to avoid magnetic saturation of the current leads if the magnetic field vector is directed towards them. One power supply is used for adjusting the vertical component and three independent supplies for the adjustment of the horizontal component. The distribution of the current in the green coils is following a “ $cos θ$ ” equivalent distribution thus improving the homogeneity. To produce a horizontal field vector, along the Y axis for instance, not only the group of coils which are in the axis are being used, but the 3 groups of green coils simultaneously. This makes it possible to increase both the field strength and the homogeneity while reducing the energy present in each coil (see Fig. 10).

Fig. 10.

Top view distribution of the power in coils for a 1.1 T magnetic field vector along the Y axis.

Table 1

Parameters of the three types of coils: vertical field main coils, vertical field shielding coils and horizontal field coils

Parameters of the coils	Blue coils	Orange coils	Green coils

	Vertical field shielding coils	Vertical field main coils	Horizontal field coils
Number of coils	2	2	12
Inner radius (mm)	240	104	58
Outer radius (mm)	250	167	81
Minimum z-position (mm)	230	60	83
Maximum z-position (mm)	250	71	218
Height (mm)	20	11	135
Thickness (mm)	10	63	23
Cross section (mm²)	200	676	3121
Number of turns	252	828	4216
Amp turns (At)	48510	159390	811580
Length (m)	389	706	1832
Intensity at 1 T (A)	192.5	192.5	192.5
Peak field on the windings (T)	1.2	2.44	5.7
Inductance of 1 group (H)	0.626		2.66
Stored Energy (kJ)	11.6		147.9
Operating temp. (K)	4.2 to 5 K following position and usage of the coil
Critical current	>440 A at 4.2 K and 6 T (manufacturer data)
Current variation speed	1 A/s ± 5% maximum
Peak field on sample area (T)	1.1 (on a 5 mm radius sphere)

Fig. 11.

Exploded representation of the three parts of the cold mass.

2.3. Mechanical design of the cold mass

The cold mass is an aluminum alloy (6082 T6) box, which contains all the coils and the bottom part of the cooling circuit. It is composed of three mains parts assembled by tied rod sized to provide mechanical cohesion during the cooling down and to withstand magnetic forces. The maximum deformation of the structure under magnetic load is 0.4 mm. The coils are mounted and bonded to the inside of cylindrical housings to ensure the best possible thermal contact. The form and positional tolerances are globally better than 1 mm (0.1 mm max by coil). The cold mass has an outside diameter of 535 mm for a height of 460 mm and a mass of 113 kg.

3. Cooling system design

The design of the cooling system is based on the use of two cryogenerators (the first one is a single stage of 100 W @ 50 K and the second is a double stage, 35 W @ 50 K and 1.5 W @ 4.2 K) used to cool the heat shields and a phase separator, starting point of the thermosiphon. The operation of the thermosiphon has been designed with a liquid conduction cooling system, in order to optimize the dynamic cooling capacity and allow a field variation as quickly as possible. This system, coupled with an external helium recovery tank (gaseous helium), allows WAVE to automatically recover in 4 hours after a “quench” (see Fig. 12). It can start completely autonomously in 8 days using only the cryogenerators which will lower the temperature of the system, then liquefy the helium and finally launch the helium circulation loop (CEA DACM Institute therm-autonom technology). A quick start is possible by injecting liquid helium directly into the system. Figure 12 shows the temperature curves of the cernox probes installed in WAVE (called C1, C2, C3, C4 and C5). The different probes localizations are: under the current lead for C1, on the top of the cold mass for C2, on the bottom of the cold mass for C3, on a green coil above the equatorial plane for C4 and on a green coil under the equatorial plane for C5 [4]

Fig. 12.

Evolution of temperature vs time, illustrating WAVE’s recovery after a quench [4].

Fig. 13.

Sample cryocooler DE-215SF from ARS with 1.36 m specific WAVE extension. On the detail picture, you can see RF connectors covered by yellow caps (for transport measurements) and the wire ribbon used for XY piezo-electric movement of the sample.

The sample chamber is a classical tube (100 mm in diameter) (see Fig. 1). It is totally decoupled and isolated from the cooling system of the magnet, which makes it possible to conveniently install a sample heating or cooling system. At the moment, a closed cycle cryocooler ARS DE-215SF is available (see Fig. 13), allowing to cool a sample down to 4 K [8]. It is equipped with 4 SMA RF 40 GHz from KMCO and XY Attocube ANP X311 movements. RF connectors is an option to use WAVE without neutron for transport measurement.

4. WAVE software

A software was also produced to (remotely) control the field and the parameters of the magnet (see Fig. 15). It records all the parameters and gives access to different types of WAVE operation (basic, normal, quench…). It allows the direction and intensity of the field to follow the rotation of the sample. This is especially important if the sample shows magnetic hysteresis. Figures 16 and 17 display measured values for the basic and normal modes.

To define the vector field we used the coordinate system $\vec{B} (ξ, η, | B |)$ where ξ represents the angle of the vector in the ${\vec{B}}_{x y}$ plane, η the angle of the vector in the ${\vec{B}}_{z}$ plane and $| B |$ the nominal value of the field (see Fig. 14).

Fig. 14.

System coordinate definitions.

Fig. 15.

Control panel of the field.

Fig. 16.

Parameter survey for a command line. (0°, 0°, $H = 0 T$ ) up to (45°, 45°, $H = 1.1 T$ ) in “basic” mode [4].

Fig. 17.

Parameter survey for the same command line as Fig. 16 but in “normal” mode [4].

Fig. 18.

Angular speed for rotation of the field [4].

Figure 16 shows a “basic” field rise up to (45°, 45°, 1.1 T). The power supplies raise the field strength independently at their programmed maximum rate to reach the set point. This procedure saves time in some cases, yet proves unable to maintain the field strength in the requested direction (the field rotates in an anarchic way). In the “normal” mode (Fig. 17), the software regulates the rising speed of each power supply in order to have a permanent control of the field orientation. This yields a nice constant increase of the field strength with the right orientation, allowing, for instance, not to disturb the samples with magnetic memory or to make measurements while the field changes. Slight differences between the power supplies in terms of connections, cable lengths, windings, quench training, affect their relative capabilities, hence the global variation rate of the field. Eddy currents are likewise an element to be taken into account in the optimization of magnetic field variations. Figure 18 shows the rotation speed as a function of the field strength (measured in 2018). These values, although they have to be further optimized, allow us above all to highlight the possibility of performing 360° controlled field rotations at 1T and in any plane.

5. Conclusion and first step on 6T2

The WAVE magnet was received mid-2017. A dedicated and optimized platform designed by the Id3D Company was built and installed by the LLB team on the neutron diffractometer 6T2.

Fig. 19.

Intensity of the (111) Bragg reflection of FeCo as a function of the in plane magnetic field.

Fig. 20.

WAVE on the LLB neutron diffractometer 6T2.

A very first neutron diffraction experiment was carried out in the beginning of 2019 at LLB. The goal was to study the magnetization at ambient temperature of a FeCo crystal in a magnetic field rotated in the horizontal plane. As expected, the intensity of selected Bragg peaks show a periodic behavior, yet the results are still being interpreted (Fig. 19). A second experiment, at low temperature, was scheduled at the end of June 2019, yet was unsuccessful because the base temperature was not cold enough on the sample.

Unfortunately, the shutdown of the Orphée reactor has put a stop to these activities. WAVE should however be re-used in other neutron facilities, but also for other purposes, like for instance transport measurements.

Footnotes

Acknowledgements

We would like to acknowledge the SigmaPhi staff, especially JL. Lancelot (SigmaPhi CEO), F. Forest (SigmaPhi Superconducting Business Unit Director), R. Pasquet (SigmaPhi Superconducting Project Leader), M. Delbecq and D. Ramauge, for their expertise and high quality work. We are also indebted to the DACM management team, who accepted this innovative technical challenge. Many people worked on the WAVE project and we shall thank F. Molinié, A. Peugeot, D. Simon, G. Authelet, L. Henrion and J. M. Gheller from DACM, as well as C. Alba-Simionesco, E. Eliot, J. L. Meuriot, S. Gautrot, J. Dupont, W. Josse, O. Tessier, C. Meunier, P. Lambert, F. Connego, and F. Legendre from LLB for their precious help. Thanks to ID3D for their innovative and efficient platform. This project would not have been possible without support and funding from CEA, CNRS, ANR (French National Research Agency) and “Région Île-de-France”. Finally, TR would like to warmly thank A. M. Bataille, the father of WAVE, for his confidence.

References

Aubert, Dispositif de génération d’un champ magnétique orientable et localement homogène, Brevet (2012), 1262070.

Aubert, Champ magnétique d’une couronne circulaire régulière d’aimants axisymétriques identique d’axes parallèles, Notes internes SACM, 2013.

Berriaud,

Aubert,

A.M.

Bataille,

Brédy,

Daël,

Klimko,

Lavie,

J.L.

Meuriot,

Nunio,

Peugeot,

J.M.

Rifflet,

Robillard and

Simon, 24th International Conference on Magnet Technology, 2015, 1 Po BC-07.

N.B.

Doan, LLB Young Researcher Day, 2018.

Petit and

A.M.

Bataille, Agence Nationale de la Recherche (2016), Project-ANR-16-CE09-0009.

Song,

Four and

Baudouy, Nucleate boiling heat transfer in a helium natural circulation loop coupled with a cryocooler, International Journal of Heat and Mass Transfer66 (2013), 64–71. doi:10.1016/j.ijheatmasstransfer.2013.07.002.

Visentin and

Baudouy, in: Helium Two-Phase Flow in a Thermosiphon Open Loop, Comsol Conference, Milan, 2009.

Wave sample cryocooler, ARS, https://www.arscryo.com/de-215.