A mechanical rotator for neutron scattering measurements

Neutron scattering is a powerful probe in condensed matter physics, which can provide valuable information about spin and lattice correlations and collective excitations in materials. The technique is used to great effect in the study of a wide range of materials: it has long been central to the exploration of magnetism and has also been transformative in our understanding of unconventional superconductors,^1–3 multiferroics,⁴ colossal magneto-resistive materials,⁵ and frustrated spin systems.^6–8 A key property of neutron scattering and one of its main strengths is that it is a weak probe of matter, interacting with samples only via the Born approximation and thus allows for reliable determination of intrinsic material properties in the bulk.⁹ This same characteristic, however, leads to stringent requirements for sample environments and must be considered in the design steps for neutron scattering instrumentation. For example, to maximize signal-to-noise ratios, care must be taken to minimize the amount of material used for equipment elements and the composition should be limited to only those elements which have small absorption or scattering cross sections.¹⁰ The weak nature of the scattering cross section also means that there is a general desire with most neutron techniques for large samples often at low temperatures, necessitating the use of large volume cryostats and carefully designed sample holders.^11–15 

These requirements present particular challenges to measurements involving the application of applied magnetic fields. To minimize cost and material in the neutron beam, general purpose magnet systems in modern facilities are almost exclusively constructed with large bore, split-coil solenoids aligned in a vertical field configuration.¹⁶ Though there are notable exceptions,^17,18 the use of horizontal field magnets in thermal neutron scattering instruments is much less common, as the simplest designs involve a horizontally aligned solenoid, unacceptably limiting the useful range of scattering angles. Similar reasoning complicates the construction of flexible vector magnet systems which can apply magnetic field in an arbitrary direction. To explore the role of applied field along different crystallographic directions, a common mode of operation is to use a vertical cryomagnet and a series of crystals cut and mounted with different horizontal scattering planes; however, this is greatly complicated in frustrated spin systems, in which field directions often have to be determined with accuracy better than 1°.^19–21 Recent years have seen neutron scattering used to explore details of quantum phase transitions and emergent particles induced by fields applied transverse to spin anisotropy directions,^22–25 but these measurements are much more difficult in high-symmetry systems, where similarly large applied fields are needed to train magnetic domains (and define global spin axes) and in the transverse configuration to drive transitions (e.g., Ref. 26).

In order to address some of these concerns, we have recently designed and constructed an insert for a vertical cryomagnet that allows in situ mechanical rotation of single crystal samples on neutron scattering beamlines. Our primary motivation was an ongoing study of the effects of applied field transverse to a training field in insulating spinel antiferromagnets, particularly Mn₃O₄,^26–28 and the insert was intended for use with a 5 T vertical magnet at helium temperatures on a triple-axis neutron spectrometer. A crucial element of the design was the driving of rotation through linear translation of the central shaft—a development which allowed for control of sample orientation without the use of electrical motors at the sample position, which are ill-suited for the high fields and radiation environment of the experiments of interest, while allowing the entire process to be motorized at a later date. Other key elements included adjustable insert length and baffle sizes, which allow the initial design to be adaptable for use in a number of different measurement conditions.

In the current paper, we lay out the design of our rotator, with extended explanations of some non-trivial features. We present preliminary neutron scattering measurements taken using the rotator on single crystalline Mn₃O₄ and discuss its performance and effectiveness. Finally, we conclude with some ongoing modifications to our design and possible future directions.

As alluded to in Sec. I, a number of environmental factors played in our rotator design. In addition to the general desire to minimize material in the beam, our design had to take into account the potential for neutron radiation to damage certain materials or produce radioactive isotopes,¹⁰ or further disrupt the operation of complex electronics.^29,30 Likewise, high magnetic fields and extremely low operating temperatures restricted our choice of materials and eliminated the use of any electronics not built to withstand such extreme operating conditions. Size constraints were another major factor: the desire for a general use piece of equipment initially favored a design that fits on the end of a standard removable insert for a top-loading cryostat, and all moving parts at the sample position and associated driving mechanisms needed to fit inside the bore of a typical neutron cryomagnet, usually considerably smaller than 50 mm in diameter. However, this proved impractical, as described below.

An overview of our final design is shown in Fig. 1, which takes the form of a purpose-built stick and baffle assembly to replace the standard sample holder assembly in a top loading cryostat. Rotation is achieved through translation of a central rod, which is coupled to a rotating sample plate through a linkage system of our own design. Translation is driven mechanically at a location exterior to the cryogenic tube and magnet, which allows us to bypass many of the restrictions laid out above. A central, and crucial, feature of the design is that rotation is achieved through mechanical linear motion of a central shaft, thus avoiding prohibitive complications associated with the use of electric motors or piezoelectrics in the presence of large magnetic fields, while at the same time allowing for immediate modification to a system with linear drive motors and automation. These features and details about mounting plate and baffle assembly are given below. Initial designs were made to optimize the use of Oak Ridge National Laboratory’s (ORNL) MAG-B vertical field cryomagnet and the HB1a fixed incident energy triple-axis spectrometer (FIETAX) at the High Flux Isotope Reactor (HFIR). This system is a 5 T symmetric-field magnet, with a 50 mm bore and 152-mm sample space. It is important to note though that efforts have been made to keep the design flexible, and the entire system is easily adapted to other vertical cryomagnets. This flexibility is also discussed below.

In construction of the insert, care was taken to minimize the amount of material associated with the holder and rotation mechanism near the sample position, and we further worked to minimize the background by using only weakly scattering materials. Due to durability, scattering intensity, and cost, the two most common materials chosen for neutron sample environment systems are alloys of aluminum or vanadium,^11,13 with the former preferred for single crystal measurements and the latter for powder diffraction experiments. Because the insert is mainly to be used to facilitate a series of single crystal scattering measurements, the lower total cross section of aluminum was preferable to the lower coherent cross section of vanadium. As such, we chose to build the portion of the machine which would be in the beam out of 6000-series aluminum alloy as much as possible. This choice had the additional advantage of simplifying the total background signal, as the beam window and sample for a typical measurement on MAG-B are both built out of a similar alloy. Background considerations also led us to seek out hydrogen-free lubricants for the interfaces between moving parts, and we ultimately selected to use Dow-Corning Molykote^® D-321 R, a high-pressure rated MoS₂ based coating, which is certified for use at extremely low temperatures.

For the portion of the insert outside the beam path, our primary material concerns were response to stray magnetic fields and durability at low temperature. Because of its intrinsic lack of magnetism and relative structural strength, we continued to use 6000-series aluminum alloy for parts nearer the sample space, particularly in moving parts, to avoid the mechanism being moved by the field. We also used the same lubricant discussed above in this space. In areas further removed from the sample space, we switched to 300-series stainless steel for its structural strength, as well as its low vapor pressure, thermal conductivity, and magnetic moment. We coupled the aluminum rotation piece to the stainless steel struts using a cylindrical copper block to act as a heat sink for the thermometry, provided by a Lake Shore Cryotronics Cernox^® CX-1030-AA negative temperature coefficient RTD. The copper block is well above the beam to minimize its contribution to the background. Figure 2 shows details of the sample mounting assembly.

The crystal is attached to a circular plate using low gauge aluminum wire threaded through several small holes. The thickness of this plate can be varied to adjust for the height of the crystal (see the figure), so that the center of rotation will be at the center of the sample. The plate is circular so that it can rotate in the table. This is necessary for sample alignment, as described below.

Figure 3 gives a schematic of the main rotation mechanism. The table with the crystal is near the center of the image. It can pivot about axle pins connecting it to the sides of the frame and is attached by a rigid strut to a horizontal yoke, which is in turn attached to a centrally mounted rod that passes through the upper part of the aluminum frame. This rod turns in the yoke through a Teflon bushing to prevent galling. The portion of the rod inside the frame is threaded to the hole it passes through. To drive rotation of the sample, the operator turns the central rod, which vertically displaces the yoke, pushing on the assembly below and moving one end of the crystal plate about the axle pins. As built, our mechanism allows rotation from −10° to 100°, where 0° is defined as the horizontal plate position.

Set screws attach the copper block to two stainless steel rods. These rods are welded to baffles partway up their length, which provide both heat shielding and extra stiffness for the overall assembly. The threaded aluminum central rod attaches to a longer stainless steel rod, and these three rods run up nearly the entire length of the stick. There, they attach to a second set of stainless steel rods which run through vacuum-tight seals in a stainless steel block at the top of the stick. The reason for the double set of rods is to make the system modular (more detail is in Sec. IV). The seals for the side rods can be loosened so that the overall length of the stick can be adjusted by several centimeters if needed. The center seal permits both rotation and translation of the rod while maintaining a vacuum tight seal—see Fig. 5 for a detailed view of the seals. This allows the rotation mechanism to be mechanically driven from the outside, where the operating environment is significantly less hostile. We have chosen to drive this rotation manually to serve as a starting point, but as rotation takes place far from the magnet position it would be a simple matter to attach an electric motor to the central rod at this point; we discuss this possibility more in Sec. IV.

A ruler on top of the steel block parallel to the rod and a circular knob with 20 notches spaced evenly around the edge attached to the rotation rod together allow the user to see the vertical displacement of the threaded screw in the rotation frame, which can be translated to the angle of rotation: large (≳1 mm) adjustments can be made using the ruler reading and small adjustments (∼0.05 mm) can be made using a vernier scale in the form of notches around the edge of the knob. The pitch of the thread in the rotation frame is chosen so that one turns the knob nine times to rotate the sample stage by 90°. However, vertical displacement and sample rotation are not linearly related. The particulars depend on the exact geometry used in the linkage system. For our dimensions

where y is the vertical displacement and θ is the angular rotation of the sample plate. We have defined y₀ to be the ruler reading at θ = 0. See also Fig. 4. Details of the feedthroughs and manipulation knob are shown in Fig. 5.

This system allows for the sample plate to be rotated over a wide range of angle about the horizontal axle pins shown in Fig. 3 but does not allow for any rotation about the perpendicular horizontal axis. Thus, care must be taken by the user to align the desired rotation axis with the horizontal axle pins. This alignment is best performed outside the sample environment. To make the process easier, we constructed the accessory shown in Fig. 6. To use it, the end of the rotation stick can be removed and attached to the aluminum riser shown, which is designed to mount on the goniometer of an alignment station of the user’s choosing, keeping the faces perpendicular to the edges and the center of rotation of the crystal at the height of the beam. By rotating the alignment plate in the table, the crystal can be aligned to the required precision and then secured by tightening the two holding screws—see Fig. 6. Alignment can be further checked by driving the rotation mechanism to θ = 90^∘ and making sure peaks in that plane are accessible. Afterward, the mechanism can be reattached to the stick, while retaining all orientational information.

As an initial test, we used the rotation insert to perform a series of measurements on a ∼100 mg single crystal of Mn₃O₄ using MAG-B and the HB1a triple-axis spectrometer at the HFIR. We initially aligned the crystal with x-rays at the University of Illinois and cut it along the tetragonal [H 0 L] plane. Figure 7 shows a picture of the crystal loaded onto the rotator in the θ = 90° configuration. Measurements were performed both with [0 0 L] and then [H 0 0] aligned along the rotation axis, shown as the vertical direction in Fig. 7. We changed between these configurations by rotating the circular plate in the sample table, as described in Sec. II and shown in Fig. 6.

For the crystal shown in Fig. 7 with [0 0 L] aligned on the rotation axis, the horizontal scattering plane corresponds to [H 0 L] when θ = 0° and [0 K L] when θ = 90°. As a first test, we mounted the samples with θ = 0° and rotated the sample 90°, and performed several measurements in the [0 K L] plane. Figure 8 shows a false color plot of neutron scattering intensity in this plane on a logarithmic scale at temperature T = 75 K and reveals several structural Bragg peaks of Mn₃O₄. This confirmed that the rotator worked as designed with better than 0.2° alignment. Also visible are powder rings associated with the (1 1 1), (0 0 2), and (2 2 0) Bragg peaks of aluminum, which allows for a direct comparison of signal intensity from our crystal to background from the unavoidable aluminum at the sample position. In most cases, Bragg peaks are seen to be 1-2 orders of magnitude more intense than the aluminum scattering. The Mn₃O₄ (0 0 4) peak specifically, which notably overlaps with the aluminum (1 1 1) ring, has an intensity nearly three orders of magnitude greater than background.

Further scattering data from Mn₃O₄ (Fig. 9) demonstrate how our rotator may be used to address some of the scientific questions mentioned in Sec. I. Mn₃O₄ is a tetragonal spinel that orders into a ferrimagnetic state with a net magnetic moment along the tetragonal [0 K 0] direction.^31–33 Recent Raman scattering studies have revealed that the ordered structure can be substantially modified with relatively small magnetic fields applied in the perpendicular direction;^26,27 however, the existence of an extensive magnetic domain structure (and thus the lack of a global b-axis) in larger samples has made it difficult to confirm and expand upon these results with neutron scattering. Making use of our rotator, we were able to define the global b-axis by cooling the sample in the presence of field using the vertical MAG-B cryomagnet, training the domains, before decreasing the field to zero and rotating the sample about a perpendicular direction. Success of this domain training protocol ([H 0 0] or [0 0 L]) can be verified using the scattering selection rules for the particular ordered state in question, which declare that magnetic peaks are at positions indexed by (1 K 0), while scattering at peaks (H 1 0) is forbidden. In Fig. 9(a), we show that one of these forbidden peaks, the (1/2 1 0), is present when cooled with H = 0, reflecting a robust domain structure, but strongly suppressed when the above procedure is applied with H = 50 kOe||b. Similarly, Fig. 9(b) shows strong intensity in (1 K 0) peaks after training, while the (H 1 0) peaks are almost entirely suppressed, demonstrating that the sample is mono-domain. Data in the latter panel were taken in the [H K 0] scattering plane, with the sample vertical now representing a true transverse field direction.

Using this setup, we have successfully explored the effect of applied magnetic field on the ordering in Mn₃O₄ for a number of different angles from the spin anisotropy direction and have been able to fully examine the field-temperature phase diagram in the true transverse field configuration.³⁴ As a further application, we have found that the rotator is useful as an additional alignment degree-of-freedom in cases where instrument goniometers are unavailable or inoperable—for example, when goniometers driven by electric motors are immobilized in the presence of large fields. With the addition of a second rotation axis to the rotator design, it may be possible to manually adjust sample orientation in situ, without the use of any instrument goniometers at all; this idea is discussed in more detail in Sec. IV.

A major concern with any design is the fidelity and reproducibility of the set angle position and the effects of mechanical backlash and slippage in the rotation couplings. In its current state, backlash effects are accounted for manually with iterative feedback from neutron scattering data. The inclusion of a sample encoder would be a simple matter and, coupled with the use of a driving motor, would undoubtedly be helpful in any future expansion of the design. A future modification to our design to incorporate such a motor is also discussed in Sec. IV.

The initial design of our rotator allows for several significant enhancements in the near future. The simplest is adaptation to other cryomagnet systems. MAG-B has an inner bore of 50 mm, the beam center is 85.7 cm from the top flange, and it uses a KF50 connector at the top of the sample space tube. These are far from universal dimensions: other neutron scattering cryomagnets at ORNL range in height up to 2 m and vary in bore down to 25 mm. Rather than build a separate insert for each magnet system, we have purposefully designed ours so that it can be modified to fit other magnets or cryostats relatively easily. For use with narrower magnets, the baffles are removable from the structural supports and can be easily replaced with any size down to 25 mm. For differing heights, the center portions of the support and rotation rods can be removed and replaced with appropriate lengths. The thermometry wire is currently attached to one of the supports so this is a slightly more complex operation, but the wire makes electrical contact with the thermometer and with the electrical feedthrough using high lead, low temperature solder, so that it can be replaced with relative ease. If frequent changes become necessary, a socket and plug system could be implemented. Finally, the KF-flanged tee at the top can be replaced with one that connects to a different size or different style connector. These replaceable parts are shown in Fig. 10.

It would also be a relatively minor modification to increase the size of the rotating sample mount. Our design currently holds a crystal up to ∼1 cm across, which is ample for elastic scattering measurements or inelastic measurements of materials with large magnetic moments. For inelastic measurements, or those involving smaller moments, larger crystals are desirable. Because the rotation frame is a wholly separate piece from the parts above it—the copper block and support and rotation drive rods—it could be replaced with a larger mechanism while leaving the rest of the setup as it is. The rotation frame is mechanically more complex than any of the other replaceable parts, but the principle is essentially the same as described above. Figure 11 shows which parts would need to be re-fabricated to accommodate larger crystals. Additionally, Eq. (1) would need to be adjusted to reflect the new rotator dimensions.

One major enhancement to the current design would be the inclusion of a motorized drive system. In addition to freeing the user from manual adjustments and allowing the system to be used in more restrictive access environments, the automation of sample rotation would open up whole new classes of experiments, such as scans of theta as an adjustment variable. With these in mind, we have designed the motorized drive system shown in Fig. 12, intended for use with a HT23-549D Nema 23 stepper motor from Applied Motion Products. This motor has been used extensively at facilities such as the Spallation Neutron Source and should be suitable for this purpose. An additional benefit of this motor is that it has an integrated encoder, which allows for automated compensation for backlash effects associated with the linear motion of the central driving rod. An additional source of backlash error will arise from the coupling between the linear driver and the rotation table, though it is our experience that these effects are minimized if translation is driven only in one direction. Care must then be taken to redefine motor zeros with respect to observed Bragg peaks before driving in the opposite direction. Further advancement of the design will require additional engineering steps.

Finally, as mentioned above, a major advantage of our design, more general than simply defining applied field direction, is that it provides an additional mechanism through which a user can orient their sample. This is useful for instruments with no or limited goniometer arc and, since the drive mechanism is far away from the sample position, the ability to adjust tilts is largely unaffected by any applied magnetic field. Furthermore, the tilt range is far greater than possible with an external goniometer, as it does not require tilting cryostats or other heavy sample environments. To take full advantage of this capability, a powerful future design modification would be the inclusion of two rotation axes, instead of one: two internal 90° arcs are a simple way of providing scattering access to 4π steradians of reciprocal space with a single sample and greatly simplify the use of complex sample environments such as cryomagnets. We are currently considering design strategies for implementing such an extension.

We have designed and constructed a mechanical rotation insert for use on low temperature cryogenic systems at neutron scattering beamlines. This system is primarily meant for performing field-direction dependent measurements on single crystals, but it could potentially be used to modify the scattering plane in situ and allow for users to study samples in multiple scattering planes in the space of a single experiment. The initial design was optimized for use with the MAG-B cryomagnet at the HFIR at ORNL; however, it is purposefully modular and can be easily adapted to a number of different top-loading cryostat systems. We have demonstrated one potential use of this system and provided data showing the successful training of magnetic domains in the ferrimagnetic spinel Mn₃O₄ and subsequent exploration of the perpendicular scattering plane. Finally, we have also laid out several directions for future development, including a system for automating the rotation and multiple axes of rotation.

This work was sponsored by the National Science Foundation, under Grant No. DMR-1455264-CAR. Experiments performed at Oak Ridge National Laboratory’s High Flux Isotope Reactor were sponsored by the Scientific User Facilities Division, Office of Basic Energy Sciences, (U.S.) Department of Energy (DOE). We would like to thank Chris Redmon and Todd Sherline of Oak Ridge National Laboratory for their feedback regarding several design components. We would also like to thank Alex Zakjevskii, Annie Farwick, and Brian Nguyen for their help in taking the measurements.