- Email: [email protected]
Construction of new spin-echo spectrometer in Jülich
Physica B 180 & 181 (1992) North-Holland
Construction
PHYSICA I!
935-937
of a new spin-echo spectrometer
in Jiilich
Michael Monkenbusch Institut fiir Festkijrperforschung,
In the neutron
The instrument
Forschungszentrum
guide hall ELLA
at the DID0
Jiilich, KFA,
reactor
Germany
FRJ2 in Jiilich a new spin-echo
spectrometer is under construction. The scattering angle range is
will reach Fourier times up to 40ns at a neutron wavelength of 0.8nm.
0.6 < 20 < 100” corresponding
to 0.08 < Q < 12 nrn-’
at A = 0.8 nm.
1. General description The spin-echo method allows for the construction of neutron spectrometers of the highest possible resolution, however they measure the intermediate scattering function, Z(Q, t), rather than S(Q, 0). The resolution depends on the maximum achievable time parameter, t, which is proportional to the maximum useful magnetic field integral of the main precession solenoids. See ref. [l] for a detailed discussion of the physical principles. The dynamic range for the time, t, may exceed a factor of 1000. These properties are especially well adapted to the study of (slow) relaxation phenomena. The first spin-echo spectrometer, built in the late 197Os, is the IN11 at the ILL, Grenoble [l, 21. The IN11 is still the prototype for the general layout of our spectrometer, however there are several improvements. The single detector is replaced by an area multidetector (32 x 32 cm, 32 x 32 cells). The use of solenoids requires precession the multidetector (length = 2 m) of larger inner diameter (35 cm). The Jiilich design has the above improvements in common with the new IN15, just starting operation at the ILL. In addition the Jiilich NSE will exhibit a maximum field integral of 0.5 T m, which is 5 times that of IN11 and 2 times IN15, with the same length of the main solenoid. To achieve reasonable neutron intensities it is essential to build the instrument as short as possible and use as large beam divergencies as possible. A new scheme to decouple the solenoids by compensating current loops at each end of the other parts of the spectrometer has been realized in the Jiilich design, allowing for a distance sample-n/2-flipper of 3.2 m. The number of useable neutrons scales with the fourth power of this distance. A three “Fresnel”-coil correction scheme has been devised and investigated for the Jiilich design, which will be able to perform the required field integral homogeneity correction. The NSE will not get an end position at an ELLAguide but will receive its neutrons (A = 0.8 nm) by a polarizing multilayer mirror on %-wafers inserted at 0921-4526/92/$05.00
W-51 70 Jidich,
0
1992 - Elsevier
Science
Publishers
an angle of 2” into the curved guide NLIIb, which feeds one of the existing small angle scattering machines (KWS2). The resulting wavelength distribution is tailored by a velocity selector (Dornier) before entering the spectrometer. A flux of 106n/cm2s (A = 0.8 nm) at the sample position is expected. 2. Precession solenoids and decoupling The design is based on hollow conductor, water cooled copper solenoids. In this case a high field integral, J,, = I IB( dl, (0.5 T m), at limited length (2 m) of the solenoid requires large power (70 kW) and a large amount of copper (2.5 t). Figure 1 displays the geometry of one of the precession solenoids. The compensation loops at both ends are electrically con-
s
j
n cc1
J_
t 400
P m
1
,O”t
i
CC3
t cc2
1
i
0 ti\ 0
I
2
3
4
x/m
Fig. 1. Field of one precession solenoid ensemble; S is the sample position, CC1 .3 denote the positions of the 3 “Fresnel”-type correction coils, representing radial current distributions j, 3(r) roughly proportional to the axial distance r. The magnetic field scale corresponds to $ of the maximum.
B.V. All rights
reserved
M. Monkenbusch
936
S
I New spin-echo spectrometer
in JzXch
n-FL
HP t
lm
1
Fig. 2. Secondary arm of the spectrometer, magnetics and main components only. The primary arm is symmetric except that the analyzer (ANA) detector (DET) group is replaced by a neutron guide to the n/2-flipper, the vflipper is present only either in front or behind the sample, S
netted in series with the central solenoid, they decouple the components of the spectrometer by fast reduction of the outer stray fields of the solenoid. This is paid by about 20% of the maximum field integral. Figure 1 shows also the magnetic vector field of the ensemble. There are zero field regions at the sample position and the position of the n/2-flipper. The nonzero but low fields required at these positions are generated by independent coils. By integration of the Bloch-equation it has been carefully checked that the field compensation is compatible with the requirement of adiabatic spin rotation along all the neutron paths. At the sample position a minimum field of 0.05 mT is needed, any larger value will unnecessarily reduce the resolution. Figure 2 shows the assembly of the spectrometer from the sample position to the detector. Due to the compensation, the external field needed for the operation of the a/2-flippers is generated exclusively by dedicated Helmholtz pairs, HP (except for the compensation of higher order corrections for off-axis positions, provided by the extra current loop, EL). The analyzer needs a field of 5 mT which is generated by a fully compensated solenoid ensemble with virtually no field at 1 m distance. The field in the sample region is controlled by some extra windings on top of the compensation loops of the main ensemble. The n-flipper operates in this low field region (B < 0.1 mT). 3. Field integral homogeneity
and resolution
The nominal resolution given in terms of the maximum time parameter, 1, as a function of field integral, Jo, and neutron wavelength, A, t--- JOA3may only be utilized if the neutron spin-echo is not yet wiped out by the effect of field integral inhomogeneity. The
maximum number of Larmor precessions in the standard operation mode of our spectrometer will be 0.4 X lo’, therefore a relative homogeneity better than lo-’ (some 10-6) is required. The solenoids of the instrument must have sufficient radial symmetry, which is ensured by a special winding scheme [3], then the inhomogeneity has only two contributions: (1) paths that are not parallel to the axis are longer than parallel paths, (2) paths in off-axis regions feel a larger field than the axial path. In ref. [3] it has been shown that both effects can be corrected by inserting three customized “Fresnel”-type coils in each precession solenoid, this result is neither affected by the compensation loops nor any other radial symmetric current loops around the beam nor by the limited validity of the approximation of the correction coil action based on Amperes law as used in ref. [3]. A full calculation without this approximation leads to an iterative scheme for the determination of the correction elements, the currents (and shapes) are resealed by several percent but the final homogeneity is of the order of lo-‘. However, the large divergency of the scattered beam hitting the multidetector leads to a noncorrected inhomogeneity of a few 10m3 compared to the ppm we aim at. Therefore the correcting current distributions (up to 50 A/mm-radius for the worst case) have to be realized with a promille accuracy. The large current densities and the requirement of neutron transparency allow only for a massive aluminum construction. The radial current distribution is shaped by a parabolic thickness variation of the circular coil together with a pseudo-spiral cut generating conducting rings of decreasing radial thickness as the radius increases. For nonzero angle between the primary and secondary spectrometer the inhomogeneity due to the influence of the “skew” magnetic field from the “other”
M. Monkenbusch
I New spin-echo spectrometer
spectrometer arm is largely reduced by the decoupling scheme, residual small corrections may be performed by current sheets.
in Jiilich
937
main parts is planned to be finished in the second half of 1992. References
4.
Conclusion
Provided that the correcting “Fresnel”-coils may be manufactured such that they realize the computed radial current distributions with an accuracy of 10m3, the NSE in Julich will be a high intensity and high resolution spin-echo spectrometer. Assembly of the
[l] F. Mezei, ed., Neutron Spin Echo Proceedings, Lecture Notes in Physics, Vol. 128 (Springer, Berlin, 1979). [2] The Yellow Book, Guide to Neutron Research Facilities at the ILL, Grenoble (1988). [3] M. Monkenbusch, Nucl. Instrum. Methods. A 287 (1990) 465.
Construction
PHYSICA I!
935-937
of a new spin-echo spectrometer
in Jiilich
Michael Monkenbusch Institut fiir Festkijrperforschung,
In the neutron
The instrument
Forschungszentrum
guide hall ELLA
at the DID0
Jiilich, KFA,
reactor
Germany
FRJ2 in Jiilich a new spin-echo
spectrometer is under construction. The scattering angle range is
will reach Fourier times up to 40ns at a neutron wavelength of 0.8nm.
0.6 < 20 < 100” corresponding
to 0.08 < Q < 12 nrn-’
at A = 0.8 nm.
1. General description The spin-echo method allows for the construction of neutron spectrometers of the highest possible resolution, however they measure the intermediate scattering function, Z(Q, t), rather than S(Q, 0). The resolution depends on the maximum achievable time parameter, t, which is proportional to the maximum useful magnetic field integral of the main precession solenoids. See ref. [l] for a detailed discussion of the physical principles. The dynamic range for the time, t, may exceed a factor of 1000. These properties are especially well adapted to the study of (slow) relaxation phenomena. The first spin-echo spectrometer, built in the late 197Os, is the IN11 at the ILL, Grenoble [l, 21. The IN11 is still the prototype for the general layout of our spectrometer, however there are several improvements. The single detector is replaced by an area multidetector (32 x 32 cm, 32 x 32 cells). The use of solenoids requires precession the multidetector (length = 2 m) of larger inner diameter (35 cm). The Jiilich design has the above improvements in common with the new IN15, just starting operation at the ILL. In addition the Jiilich NSE will exhibit a maximum field integral of 0.5 T m, which is 5 times that of IN11 and 2 times IN15, with the same length of the main solenoid. To achieve reasonable neutron intensities it is essential to build the instrument as short as possible and use as large beam divergencies as possible. A new scheme to decouple the solenoids by compensating current loops at each end of the other parts of the spectrometer has been realized in the Jiilich design, allowing for a distance sample-n/2-flipper of 3.2 m. The number of useable neutrons scales with the fourth power of this distance. A three “Fresnel”-coil correction scheme has been devised and investigated for the Jiilich design, which will be able to perform the required field integral homogeneity correction. The NSE will not get an end position at an ELLAguide but will receive its neutrons (A = 0.8 nm) by a polarizing multilayer mirror on %-wafers inserted at 0921-4526/92/$05.00
W-51 70 Jidich,
0
1992 - Elsevier
Science
Publishers
an angle of 2” into the curved guide NLIIb, which feeds one of the existing small angle scattering machines (KWS2). The resulting wavelength distribution is tailored by a velocity selector (Dornier) before entering the spectrometer. A flux of 106n/cm2s (A = 0.8 nm) at the sample position is expected. 2. Precession solenoids and decoupling The design is based on hollow conductor, water cooled copper solenoids. In this case a high field integral, J,, = I IB( dl, (0.5 T m), at limited length (2 m) of the solenoid requires large power (70 kW) and a large amount of copper (2.5 t). Figure 1 displays the geometry of one of the precession solenoids. The compensation loops at both ends are electrically con-
s
j
n cc1
J_
t 400
P m
1
,O”t
i
CC3
t cc2
1
i
0 ti\ 0
I
2
3
4
x/m
Fig. 1. Field of one precession solenoid ensemble; S is the sample position, CC1 .3 denote the positions of the 3 “Fresnel”-type correction coils, representing radial current distributions j, 3(r) roughly proportional to the axial distance r. The magnetic field scale corresponds to $ of the maximum.
B.V. All rights
reserved
M. Monkenbusch
936
S
I New spin-echo spectrometer
in JzXch
n-FL
HP t
lm
1
Fig. 2. Secondary arm of the spectrometer, magnetics and main components only. The primary arm is symmetric except that the analyzer (ANA) detector (DET) group is replaced by a neutron guide to the n/2-flipper, the vflipper is present only either in front or behind the sample, S
netted in series with the central solenoid, they decouple the components of the spectrometer by fast reduction of the outer stray fields of the solenoid. This is paid by about 20% of the maximum field integral. Figure 1 shows also the magnetic vector field of the ensemble. There are zero field regions at the sample position and the position of the n/2-flipper. The nonzero but low fields required at these positions are generated by independent coils. By integration of the Bloch-equation it has been carefully checked that the field compensation is compatible with the requirement of adiabatic spin rotation along all the neutron paths. At the sample position a minimum field of 0.05 mT is needed, any larger value will unnecessarily reduce the resolution. Figure 2 shows the assembly of the spectrometer from the sample position to the detector. Due to the compensation, the external field needed for the operation of the a/2-flippers is generated exclusively by dedicated Helmholtz pairs, HP (except for the compensation of higher order corrections for off-axis positions, provided by the extra current loop, EL). The analyzer needs a field of 5 mT which is generated by a fully compensated solenoid ensemble with virtually no field at 1 m distance. The field in the sample region is controlled by some extra windings on top of the compensation loops of the main ensemble. The n-flipper operates in this low field region (B < 0.1 mT). 3. Field integral homogeneity
and resolution
The nominal resolution given in terms of the maximum time parameter, 1, as a function of field integral, Jo, and neutron wavelength, A, t--- JOA3may only be utilized if the neutron spin-echo is not yet wiped out by the effect of field integral inhomogeneity. The
maximum number of Larmor precessions in the standard operation mode of our spectrometer will be 0.4 X lo’, therefore a relative homogeneity better than lo-’ (some 10-6) is required. The solenoids of the instrument must have sufficient radial symmetry, which is ensured by a special winding scheme [3], then the inhomogeneity has only two contributions: (1) paths that are not parallel to the axis are longer than parallel paths, (2) paths in off-axis regions feel a larger field than the axial path. In ref. [3] it has been shown that both effects can be corrected by inserting three customized “Fresnel”-type coils in each precession solenoid, this result is neither affected by the compensation loops nor any other radial symmetric current loops around the beam nor by the limited validity of the approximation of the correction coil action based on Amperes law as used in ref. [3]. A full calculation without this approximation leads to an iterative scheme for the determination of the correction elements, the currents (and shapes) are resealed by several percent but the final homogeneity is of the order of lo-‘. However, the large divergency of the scattered beam hitting the multidetector leads to a noncorrected inhomogeneity of a few 10m3 compared to the ppm we aim at. Therefore the correcting current distributions (up to 50 A/mm-radius for the worst case) have to be realized with a promille accuracy. The large current densities and the requirement of neutron transparency allow only for a massive aluminum construction. The radial current distribution is shaped by a parabolic thickness variation of the circular coil together with a pseudo-spiral cut generating conducting rings of decreasing radial thickness as the radius increases. For nonzero angle between the primary and secondary spectrometer the inhomogeneity due to the influence of the “skew” magnetic field from the “other”
M. Monkenbusch
I New spin-echo spectrometer
spectrometer arm is largely reduced by the decoupling scheme, residual small corrections may be performed by current sheets.
in Jiilich
937
main parts is planned to be finished in the second half of 1992. References
4.
Conclusion
Provided that the correcting “Fresnel”-coils may be manufactured such that they realize the computed radial current distributions with an accuracy of 10m3, the NSE in Julich will be a high intensity and high resolution spin-echo spectrometer. Assembly of the
[l] F. Mezei, ed., Neutron Spin Echo Proceedings, Lecture Notes in Physics, Vol. 128 (Springer, Berlin, 1979). [2] The Yellow Book, Guide to Neutron Research Facilities at the ILL, Grenoble (1988). [3] M. Monkenbusch, Nucl. Instrum. Methods. A 287 (1990) 465.