Electrostatic levitation facility optimized for neutron diffraction studies of high temperature liquids at a spallation neutron source

Information about atomic structure is the key to understanding and controlling many of the physical properties in a wide range of materials. X-ray and neutron scattering techniques allow for the study of short- and medium-range atomic order and dynamics in liquids^1–4 and have provided valuable insight into the connection between structure and thermophysical properties,^5–7 metastable phase formation,¹ and phase transformations.⁸ At very high temperatures, the use of conventional furnaces for scattering experiments on liquids is problematic; sample reactivity with the container and heterogeneous nucleation both limit access to the supercooled liquid state. Further, induced ordering in the liquid due to contact with a substrate has also been observed, although this ordering, which induces structures commensurate with the container surface structures, does not tend to persist very deep into the volume.^9,10 Finally, scattering artifacts introduced into the measured signal due to the interactions of the probe beam with the container and secondary scattering from the container greatly complicate data analysis. To alleviate these problems, the use of levitation techniques for the study of the structure and dynamics in liquids has grown over the last two decades. In particular, the development of facilities that utilize electromagnetic^11–13 and electrostatic^14–17 levitations for synchrotron X-ray scattering studies have played a key role in advancing knowledge of the structural evolution in metallic liquids and glasses. With the recent construction of the Spallation Neutron Source (SNS) user facility at Oak Ridge National Laboratory (ORNL), the opportunity exists to utilize the high neutron flux and time-of-flight techniques to extend investigations of atomic structure and dynamics.

Synchrotron radiation-based techniques, such as diffuse X-ray diffraction,^1,6 anomalous X-ray diffraction,¹⁸ and X-ray absorption,¹⁹ have all been employed for studies on levitated samples, but have limitations for studies of metallic liquids. Neutron scattering is a particularly powerful technique for complementary studies in condensed matter due to its ability to probe short- and medium-range order, in addition to dynamics and magnetism. Further, the commercial availability of nuclear isotopes of many elements allows quantitative probes of chemical and topological ordering in many metallic alloys using systematic isotopic substitution. As examples, the Cu–Zr,^4,20 Ni–B,²¹ Ni–Nb,^21–24 and TiZrNi^25,26 alloys are each reported, from simulation or previous neutron scattering studies, to display ordering on length scales not present in analogous x-ray studies. For these liquids, the combination of contrast invariance between the X-ray scattering strengths of the constituents²⁷ and energy dependent sample absorption hinder the ability to adequately determine the structural character of the liquid. While in most cases, X-ray and neutron scattering techniques are complementary, the increased complexity and cost inherent in neutron sample environments appropriate for metallic alloys, semiconductors, and ceramics have made them slow to develop.

In the present work, an electrostatic levitation facility (NESL, for Neutron ElectroStatic Levitator) for use with neutron production user facilities is described and commissioning results are presented. For weakly scattering materials such as metallic liquids, electrostatic levitation is the preferred technique, allowing the sample to be kept in the supercooled liquid state for significant amount of time and providing a uniform sample temperature. It provides a cleaner environment for scattering studies than competing techniques such as aerodynamic and electromagnetic levitation. Further, unlike electromagnetic levitation, the technique is not limited to conductive materials. The apparatus is designed for integration into the Nanoscale-Ordered Materials Diffractometer (NOMAD), Wide Angular-Range Chopper Spectrometer (ARCS), and Cold Neutron Chopper Spectrometer (CNCS) beamlines at the SNS to utilize the high flux and excellent detector coverage around the sample. Key design features of NESL are discussed here, including fundamental aspects of electrostatic levitation and the unique design elements necessary to overcome the challenges associated with neutron investigations at a user facility. As a demonstration of the quality of data that can be obtained with NESL, the total structure factor for Zr₆₄Ni₃₆, measured from elastic neutron scattering studies made at a temperature that is slightly above the liquidus temperature, is presented.

The key design features are described in this section. Treated as sub-units, they are discussed separately. A schematic exploded view of the NESL is shown in Fig. 1. A unit called the “keystone assembly” (labelled as “A” in Fig. 1) contains the sample and is inserted into the vacuum vessel (labelled as “B” in Fig. 1) for processing. The vacuum vessel interfaces directly with the scattering beam line.

Levitation and sample processing for several hours at a high temperature requires a high vacuum (10⁻⁶ − 10⁻⁷ Torr) inside the NESL. Schematic views of the entire NESL apparatus are shown in Fig. 2. The vacuum vessel used to maintain the high vacuum and which interfaces directly with the NOMAD detector chamber is shown in Fig. 2(a). A large 37 in. diameter stainless steel flange (labelled as “A” in Fig. 2) connects via an O-ring seal to a stainless steel full nipple called the “headpiece” (labelled as “B” in Fig. 2) which itself is connected via a 16.5 in. Conflat® (Varian, Inc., Santa Clara, CA) (CF) flange to another full nipple called the “tailpiece” (labelled as “C” in Fig. 2). The tailpiece is positioned to allow the neutron beam to travel through the diameter and scatter from the levitated sample. Because the detector chamber (external to the NESL) is operated under vacuum (∼10⁵-10⁻⁶ Torr), the large 37 in. flange has a smooth O-ring sealing surface (nominally a 33 in. diameter circle that is concentric with the central axis of the chamber) on its bottom side to allow mating with the Tank-Reducer-D32, following SNS standard sample environment interface specifications.²⁸ The tailpiece is essentially an aluminum full-nipple cylinder with removable 1/16 in. thick aluminum scattering windows (labelled as “D” in Fig. 2), which wraps azimuthally around the central axis in a geometry consistent with detector coverage. A more detailed description of the tailpiece is given in Section II F. A cutaway view of the keystone assembly and the vacuum vessel is shown in Fig. 2(b). The electrode assembly, that controls levitation (labelled as “E” in Fig. 2 and described in detail in Section II C), is mounted to an instrumentation entry port called the “keystone” (labelled as “I” in Fig. 2 and described in detail in Section II B), which includes an oil-free turbo molecular vacuum pumping system (Pfeiffer Vacuum GmbH, HiPace® 700) (labelled as “M” in Fig. 2). The keystone moves vertically and sample access is provided through an O-ring sealed breakpoint (labelled as “F” in Fig. 2). The assembly in Fig. 2 allows independent control of the vacuum levels in the detector chamber and the NESL sample environment.

The functions of the instrumentation keystone are to hold devices for levitation and sample processing in place, allow precise adjustments and allow optical and electrical access through flanges. In electrostatic levitation, the three-dimensional position of the sample is determined by using two collimated light sources placed in orthogonal orientations to illuminate the sample; the sample shadow is then measured using two orthogonal position-sensitive detectors (PSDs). In NESL, the light sources are two expanded helium–neon (He/Ne) lasers. The laser beams are passed through the electrode assembly in order to determine the Cartesian coordinates of the sample from the PSD signals. Deviations from the set position are used to construct an error signal for a control algorithm that actively adjusts the voltages on three sets of nearly orthogonal electrodes to maintain the sample position. Details of the control algorithm and its implementation are provided elsewhere.^15,29 A detailed discussion of the levitation implementation in NESL is provided in Section II C.

In order to optimally utilize the detector coverage in NOMAD and other beam lines and to minimize elements exposed to the neutron beam, most of the optical signals are directed onto the sample from above or below the horizontal sample plane using an assembly of mirrors. The keystone is raised and lowered using the lifting plate (labelled as “K” in Fig. 2). In addition to the vacuum breakpoint where the keystone rises (labelled as “F” in Fig. 2), there is an additional O-ring breakpoint labelled as “J” in Fig. 2. In one configuration, the lifting plate can be used to remove the entire keystone, associated instrumentation and the electrode assembly; in a second configuration, the lifting plate is used to remove only the electrode assembly, leaving the keystone in place. A connecting tube (labelled as “H” in Fig. 2) is used to connect the electrode assembly and the instrumentation feedthrough (labelled as “L” in Fig. 2) directly below the turbo pump.

Twelve 2.75 CF flanges are welded into the keystone (two are labelled as “G” in Fig. 2) to allow instrumentation to be mounted to and directed into the vacuum chamber. The top and bottom mirror platforms are labelled as “N” and “O,” respectively, in Fig. 2 and the electrode assembly is labelled as “E.” One of the optical paths is indicated in Fig. 2(c). Optical instruments mounted to the keystone utilize paths such as the one indicated to probe the sample. A top down view of the keystone is shown in Fig. 3(a), with the instrumentation location assignments indicated. Optical instruments (such as the heating and positioning lasers, pyrometer, position-sensitive detectors, and cameras) are mounted above 2.75 CF vacuum flanges that contain quartz or Corning 7056 glass windows. The instrumentation is discussed Subsections II C–II F in substantial detail.

The mirror platform, shown in Fig. 3(b), is critical to the operation of the NESL. It provides sample access to non-contact instrumentation while also allowing large solid angle detector coverage of the scattering from the sample. Inexpensive silver coated Borofloat® 33 mirrors (Newport λ/10 @ 632.8 nm; model 10SD520ER.2) are used in the facility (labelled as “D” in Fig. 3(b)) to direct all non-contact optical instrumentation to and from the sample. All of the mirror mounts are custom fabricated to be vacuum compatible with the NESL operation.

The electrode assembly is shown in Fig. 4. In NESL, the samples (mass typically 100 mg to 500 mg) acquire a positive surface charge when placed in the strong vertical electric field (0–3000 V/mm). The resulting electrical force on the sample balances the gravitational force, causing the samples to levitate. Sample positioning in the horizontal plane is maintained with a much smaller field (nominally 0 V/mm, but with a maximum of approximately 500 V/mm in magnitude). The top electrode (labelled as “A” in Fig. 4) has an approximately 30 mm radius of curvature, which creates an additional lateral electric field contribution that enhances stability. Both the top and bottom electrodes (bottom electrode labelled as “B” in Fig. 4) have approximately 8 mm diameter holes (labelled as “E” in Fig. 4; the 8 mm diameter hole in the top electrode is not visible in the view) that are axially bored through them. The hole in the top electrode allows samples to be changed, while the hole in the bottom electrode allows a sample post to be raised for sample changing or sample processing to evaporate impurities prior to levitation. The bottom portion of the electrode assembly is mounted in a curved sample retrieval apparatus (labelled as “D” in Fig. 4), designed to enhance experimental efficiency by recycling samples that may become unstable during processing via the sample post.

Two He/Ne lasers (CVI Melles Griot, model 25-LHP-925) are mounted to the keystone, their outputs are fiber coupled, expanded 20 times (CVI Melles Griot, model 09 LBX 005), and then passed through the electrode assembly on lines defined by pairs of mirrors. The lateral electrodes (labelled as “C” in Fig. 4) are shaped to allow the expanded beams (∼10 mm diameter) to pass through the electrode assembly while each still capturing a complete shadow of the sample. The silhouette of the sample is passed to two silicon position sensitive detectors (PSDs) (On-Trak Photonics, PSD: model PSM 2-20; amplifier: model OT-301DL), which are mounted to the keystone (each diametrically opposite to a laser) and positioned using compact stepper-controlled platforms (Newport Corporation, two-axis platform: model UMR5XY-TRA12PPD; controller/driver: model SMC100PP). When the laser beams pass through the center of the electrode assembly, the position and velocity of the silhouette of the sample within the beam as captured on the PSDs are interpreted by a sophisticated gain-scheduling feedback algorithm²⁹ and the electrode voltages are modified to keep the sample stationary. High-slew-rate, high-voltage, amplifiers (vertical high voltage amplifier: Trek, Inc., model P0621N; lateral high voltage amplifier: Matsusada Precision, Inc., model AMS-5B6) are used to control these voltages. The vertical high-voltage amplifier has a maximum output of 30 kV (negative relative to the bottom electrode), producing a maximum electric field strength of approximately 2.5 kV/mm. Operating conditions are typically 50%-75% of these values. The positional stability was measured to be approximately ±15 μm in both the vertical and lateral directions for solid as well as liquid samples. Additionally, samples have been processed in the supercooled liquid state for several hours at a time. Processing time is practically limited by impurities in the sample and vacuum level rather than inherent limitations in the levitation system. Solid scattering calibration standards, discussed in Section III, have been successfully levitated uninterrupted for 8 h while at ambient temperature.

The processing of liquids in the NESL is made particularly difficult by neutron activation and sample evaporation. The metallic alloys studied can become activated when exposed to neutrons of energies accessible at high intensity beamlines, such as NOMAD at the SNS. This can make them potentially highly radioactive, requiring that they be carefully removed after processing and stored until they decay to a safe level. A more serious concern, however, is the sample evaporation during processing. Metallic liquids in a vacuum evaporate at a rate that is described by the Langmuir equation³⁰ (when properly corrected for the activity of the constituents). The evaporation rate of the liquid alloy constituents can vary dramatically at the same temperature (by many orders of magnitude), depending on the vapor pressure of the elements. Components in the NESL vacuum chamber in proximity to the liquid sample will develop a deposition layer of potentially neutron-activated material and, as such, engineering steps have been implemented to mitigate this. Due to their close proximity to the sample, the levitation electrodes receive the most deposition. Therefore, all six electrodes are designed to be cheap, easy to manufacture (most experiments in NOMAD can be carried out with electrodes machined out of aluminum alloy 6061), and easily and quickly replaced. The lateral and bottom electrodes are fixed to the sample collector (labelled as “D” in Fig. 4), which can itself accumulate activated deposition. The sample collector is removable and held in place using a three-point kinematic-type mount, allowing it to also be easily replaced.

Preferential evaporation of high vapor pressure constituents can also cause the sample composition to change with processing time. In NESL, a sample will often be processed until that change exceeds a predetermined tolerance level, at which time the sample is solidified and discarded by lowering the sample post and dropping the sample through the bottom electrode into a storage collector at the bottom of the chamber. Another sample is then introduced by raising the post into the top electrode and dropping a new sample from the sample carousel. Up to 30 samples can be loaded into the carousel at a time, greatly improving the efficiency of operation of the NESL since evacuation times for the chamber from atmospheric pressure to the level suitable for processing can require several hours.

Samples in NESL are heated using two fiber-coupled diode lasers of wavelength 980 nm (nLight Corporation Pearl™ Fiber Coupled Diode Laser, model series P4). The lasers are directed into the chamber using collimators mounted on the keystone (see Fig. 3(a)) and then onto the sample from mirrors below at nearly opposite azimuthal angles for enhanced temperature homogeneity. The lasers are each capable of 110 W continuous output (driven by VueMetrix, model MV-40 Laser Controllers) and are actively cooled using thermoelectric cooling units (VueMetrix, model Vue-TEC24 Temperature Control System). The heating lasers are expanded to a diameter of approximately 6 mm at the sample position by defocusing the collimation optics (Micro Laser Systems, Inc., FC10-IR-T/SMA fiber collimators). All heating lasers are interlocked and the NESL sample environment is engineered as a class 1 laser enclosure as per ANSI laser safety standards.

The temperature of the liquid samples is measured between 500 °C and 2000 °C using a Process Sensors Metis MQ22 two-color ratio pyrometer, operating at 1.45 and 1.80 μm wavelengths. The absolute temperature scale is calibrated by matching the end of the melt plateau in the time/temperature curve obtained on heating in the NESL to the largest endothermic transition signature measured in a differential thermal analyzer.³¹ The mirror used to direct the pyrometer view onto the sample also receives a sample deposition layer over time, however, that deposition accumulates very slowly due to the large distance between the sample and the mirror. If deposition were to build up quickly on the mirror, the measured temperature would begin to deviate from the actual sample temperature. By monitoring the required average heating laser power necessary to maintain the sample temperature, we obtain a rough indicator for deposition level on the mirrors and can determine when it is necessary to replace the mirrors. Generally speaking, mirror replacement is infrequent as deposition is very slow to accumulate and does not appreciably affect pyrometer temperature within a single sample.

Neutron diffraction studies of metallic liquids require extended acquisition times because the diffuse scattering is spread over a wide range of angles rather than Bragg scattering from crystalline solids. In order to acquire sufficient statistics, the NESL is normally operated in an isothermal mode. To control the sample temperature for several hours at a time, an automated proportional-integral (PI) control is implemented using the pyrometer reading as an error signal and controlling the input currents of the lasers. The controller is well tuned, displaying excellent temperature stability over short (seconds) and long (hours) timescales. An example of the excellent thermal stability for a NESL sample is shown in Fig. 5.

During the heating of the levitated sample, hydrocarbons and adsorbed gases evaporate from the sample surface reducing the net surface charge on the sample. Since the upper range of the high voltage is limited, this charge loss must be mitigated. To counteract this process, the surface charge is actively maintained with a high-intensity ultraviolet light source. The photoelectric effect causes the ejection of electrons from the sample surface, thereby increasing the surface charge. A helium discharge lamp (Omicron Nanotechnology, model HIS 13) generates a high flux of high-energy photons (∼21 eV) that are directed onto the levitated sample with a capillary tube. The capillary, extending from the UV discharge chamber, ends approximately 60 mm from the sample location. The discharge chamber is connected directly to the NESL and helium gas flows through the discharge at a pressure of approximately 10⁻¹ Torr. The UV lamp is differentially pumped in two distinct stages so as to not degrade the vacuum in the main chamber of NESL (10⁻⁶ − 10⁻⁷ Torr).

The restrictive NOMAD well geometry requires that the tailpiece, to which the UV lamp is mounted, has a complicated design. Figure 6 shows the tailpiece and the UV lamp connection port (labelled as “A”). The lamp is water cooled, has multiple vacuum and gas lines, and has a high voltage (∼1 kV) connection for the discharge point. All of those connections (not shown in Fig. 6) are made using vacuum compatible connections through the large 37 in. stainless steel flange (labelled as “A” in Fig. 2). A large section of the bottom mirror platform (see Fig. 3(b)) has to be removed to allow the insertion of the UV lamp.

The primary purpose of the NESL tailpiece (Fig. 6) is to allow a large scattering solid angle from the sample. The aluminum scattering windows (labelled “B” in Fig. 6) wrap azimuthally around the axis of the chamber and are removable via an O-ring connection. The neutron beam enters the tailpiece at point “D” in Fig. 6, traverses the chamber along a diameter, and exits through an identical port not visible in the view. Figure 7 shows the details of the neutron scattering optics, as configured for the NOMAD beamline. Extensive optics are required to reduce background scattering of the primary beam from the entrance and exit vanadium windows, the entrance and exit ports, and the internal components, including the electrode assembly. The NESL scattering tailpiece and internal components allow scattering at angles approximately 30° above and below the sample horizontal plane and complete azimuthal scattering except for approximately ±20° downstream from the sample and approximately ±18° upstream of the sample. The upstream and downstream solid angles are inaccessible to neutron detectors by tailpiece structural supports, internal components, and neutron collimation optics.

In this section, we demonstrate the feasibility of a neutron diffraction study of a metallic liquid near its melting temperature in a very short counting time (∼30 min) using a very small sample (∼100 mg). Samples of Zr₆₄Ni₃₆ were prepared by arc melting high purity Zr (99.95 at. %, including Hf nominal 3 at. %) and Ni (99.995 at. %) on a water-cooled Cu hearth in a high purity (99.999 at. %) Ar atmosphere. Details of the sample preparation can be found elsewhere.³² After an initial processing from room temperature, isothermal scattering experiments were conducted at several temperatures. The integrated intensity data obtained from the levitated liquid sample at 1039 °C after a 30 min exposure are shown in Fig. 8(a). Both a background exposure (no sample and a 1 h exposure) and a vanadium normalization measurement (levitated vanadium sample of mass 386.8 mg for a 1 h exposure) were made in order to convert the raw scattered intensity data obtained from the sample to the total structure factor, S(q) (shown in Fig. 8(b)). A standard vanadium sample is used to calibrate the detector signal because it is nearly a completely incoherent scatterer. The measured diffraction data allow resolution corrections in the static structure factor as well as normalization on an absolute scale because the scattering cross section is well known. The details of the correction can be found elsewhere.^33,34 The static structure factor is truncated at q = 15 Å⁻¹ in Fig. 8 because no significant structural features appear at higher momentum transfer; however, NOMAD is capable of acquiring data to a momentum transfer well above q = 30 Å⁻¹. The static structure factor oscillates well around unity, displays excellent statistics for such a small sample and short acquisition time and the shape of the structure factor, and the positions of the main peak and a pre-peak are consistent with previous neutron scattering studies of Zr₆₄Ni₃₆.⁵ These characteristics demonstrate the high quality of the data and the effectiveness of the applied corrections.

In summary, the design of a novel electrostatic levitation facility optimized for neutron scattering studies of liquids was presented. The facility is designed to enable efficient experimentation, allowing multiple samples to be loaded without breaking vacuum, is capable of long experimental runs, and providing fast and economical management of neutron-activated sample deposition within the chamber. The scattering optics have been designed to minimize the scattering background, particularly important for weakly scattering liquids. The apparatus has been successfully deployed on the NOMAD at the SNS to utilize the high flux and excellent detector coverage around the sample. Initial scattering results from liquid Zr₆₄Ni₃₆ near its melting temperature have been presented and the high quality of the total structure factor indicates an effective background subtraction and low secondary scattering from the chamber. Since the preparation of this work, multiple successful scattering experiments have been conducted including those on deeply supercooled metallic liquids.

The design, construction and testing of NESL were supported by the National Science Foundation under Grant No. DMR-0959465. The work at Iowa State University was supported by the National Science Foundation under Grant No. DMR-1308099. A.K. acknowledges partial support from the U.S. Department of Energy, Office of Basic Energy Science, Division of Materials Sciences and Engineering under Contract No. DE-AC02-07CH11358. Use of the Spallation Neutron Source at Oak Ridge National Laboratory was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Scientific User Facilities Division. Help from Tony Biondo, Justin Carmichael, John Carruth, Cory Fletcher, Todd Hardt, Kenneth Herwig, Denny Huelsman, Mark Loguillo, Mark Rennich, and Harley Skorpenske is gratefully acknowledged. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the NSF.