New generation high performance in situ polarized He system for time-of-flight beam at spallation sources

C. Y. Jiang, X. Tong, D. R. Brown, A. Glavic, H. Ambaye, R. Goyette, M. Hoffmann, A. A. Parizzi, L. Robertson, and V. Lauter 1Instrument and Source Division, Neutron Sciences Directorate, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6393, USA 2Quantum Condensed Matter Division, Neutron Sciences Directorate, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6393, USA 3Laboratory for Neutron Scattering and Imaging, Paul Scherrer Institute PSI, Villigen, Switzerland


I. INTRODUCTION
2][3][4][5] In particular, recently the in situ 3 He system based on spin-exchange optical pumping (SEOP) which enables continuous pumping of 3 He on a neutron instrument has gained interest among the neutron scattering community [6][7][8][9][10] because it is capable of overcoming the decay of 3 He polarization present in a conventional ex situ pumped system.At Oak Ridge National Laboratory (ORNL), the first in situ system was commissioned in 2012 as a neutron polarization analyzer at the Magnetism Reflectometer (MR) 6 at the Spallation Neutron Source (SNS), which achieved 76% 3 He polarization.This system gave an average analyzing efficiency of 98% and 25% polarized neutron transmission from 2 to 5 Å wavelength band, for neutrons that are spin parallel to the 3 He.The overall size of the system, particularly the length (100 cm) along the neutron path, requires a sufficient space thus limits the use on other neutron instruments at ORNL.Later, a more compact 2nd generation in situ system (50 cm along the neutron path) was developed for the triple-axis spectrometer HB3 as a polarizer at the High Flux Isotope Reactor (HFIR) with 71% 3 He polarization achieved. 7Compared to the previous version, the 2nd generation system has a simplified optical layout and a smaller solenoid while still retaining the size of the maximally usable 3 He cell.Encouraged by the results, a new generation a) tongx@ornl.gov of the high performance in situ polarized 3 He system for the time-of-flight (TOF) beam at the MR was developed.The 3 He polarization produced with the new optical system attained 84% and was maintained for 3 weeks.High 3 He polarization resulted in the transmission function varying from 50% to 15% for the polarized neutron beam with the wavelength band of 2-8 Å.Here we report the design of the new optical system, construction, parameters, and neutron experimental data obtained with the new assembly.A distinct aspect of this system lies in its ability to be effectively used in the time-offlight instruments with a broad wavelength band, which was not available until now.

II. 3 He CELL OPTIMIZATION FOR BROAD WAVELENGTH BAND NEUTRONS
The main challenge of using the in situ polarized 3 He system for time-of-flight (TOF) beam is achieving high neutron polarization and transmission over a broad wavelength band, and high 3 He polarization is the essential key to this goal.The polarization and transmission for an unpolarized neutron beam passing through a polarized 3 He cell can be described by the following two equations: T n (λ) = T e exp(−nσ 0 lλ) cosh(nσ 0 lλP He ) = T e exp(−O) cosh(OP He ), where T e is the transmission through 3 He cell windows (T e = 0.87 is assumed for all calculations in this section 11 ), P He is the 3 He polarization, and O is the opacity of the cell and has the relation of where n is the number density of 3 He atoms, l is the cell length, σ 0 is the absorption cross section for 1 Å wavelength neutron, and λ is the neutron wavelength.A trade-off between P n and T n occurs naturally and an optimization of n * l (proportional to 3 He cell pressure * length) is strongly dependent on the experimental conditions and requirements.Such an optimization is usually done by maximizing the figure of merit (FOM) defined by the following equation: Thus, the opacity O that maximizes the FOM can be calculated depending on the 3 He polarization and is given by the following equation: 1 Figure 1 shows a calculated behavior of O(P He ) as a function of P He , unveiling that the opacity stays relatively flat around O = 2 for P He < 80% with a breakthrough increase at P He > 80%.
A broad local minimum with O min = 1.865 occurs around P He = 55%.Using Eq. ( 5), the optimization of n * l can be easily achieved for monochromatic neutron beam but becomes elusive for a neutron beam with broad wavelength band.To overcome this problem, we propose to use the following scheme for the optimization.First the n * l is calculated based on O = 2 at a specific wavelength λ 0 within the band, then assuming a representative 3 He polarization of 80%, the figure of merit FOM av averaged over the wavelength band is calculated as follows: where λ min and λ max correspond to the minimum and maximum of the wavelength band and s(λ i ) represents the weight  of neutrons at wavelength λ i , which depends on the neutron spectrum.The λ 0 that maximizes the value of FOM av (λ 0 ) will then be used for the cell thickness optimization.An example is shown as following: Assuming a uniformly distributed neutron spectrum with wavelength band between 2 and 8 Å, Figure 2 shows FOM av (λ 0 ) as a function of wavelength, which maximizes at about 5.4 Å.Thus, the cell thickness will be optimized at 5.4 Å so that nσ 0 l * 5.4 = 2.
It is vital to achieve high 3 He polarization for good performance of the polarizer, as demonstrated below.For the wavelength band mentioned above and optimized 3 He cell thickness at 5.4 Å nσ 0 l = 2 5.4 = 0.37 , a series of plots of neutron transmission (T) and polarization (P) vs. 3 He polarization (from 60% to 85% in 5% increment) are shown in Figure 3 which clearly demonstrates increasing 3 He polarization improves both the polarization and transmission for all neutron wavelengths, but more dramatically higher 3 He polarization increases neutron transmission at longer wavelengths and neutron polarization at shorter wavelengths.FIG. 3. The neutron polarization (P) and transmission (T) as functions of the neutron wavelength for different 3 He polarizations assuming nσ 0 l = 0.37.
The FOM used in these considerations is a good measure for experiments where the spin-flip is comparable to the nonspin-flip signal.However, if the intensities of the signals of the spin-flip and non-spin-flip channels are different by several orders of magnitude, the spin-flip signal is extremely difficult to measure even with a good neutron transmission through the analyzer.The measurement requires a very high neutron polarization with 1-P being comparable to the signal ratio.In such a case, it is advantageous to choose a better average polarization and tolerate the increased counting times needed due to the lower transmission, as is common in reflectometry and off-specular scattering measurements.

III. DESIGN AND CONSTRUCTION
The goal of the in situ 3 He system is to provide a high performance 3 He analyzer for the MR beamline with high and stable 3 He polarization.Development of high performance features such as the laser safety interlock, real-time 3 He polarization monitoring, NMR flipping, and motorized movement of the system are essential for effective operation during experiments.
A schematic picture of the system is shown in Figure 4.The outer frame in Figure 4(a) is manufactured from stainless steel with sections welded together to provide a rigid and sturdy structure.The system is enclosed using laser safety panels (Figure 4(b)), with all but one side being attached with security Torx screws to prevent unauthorized access as well as limiting the need for interlock switches on every panel of the enclosure.The front main access panel has two interlock switches that will shut down the laser system and lock it out of operation if the panel is detached.All connections pass through bulkhead connectors, including the heated air connections for the oven and the cooling water that keeps the µ-metal within its operating temperature range.The enclosure also incorporates cooling fans that maintain airflow inside the assembly and serves to maintain a reasonable temperature (below 50 • C) for the optics and other electronics associated with the laser system.A heater box is mounted on top of the frame and supplies heated compressed air into the 3 He cell oven.The heater is interlocked with the airflow for safe operation at the beamline.Both vertical side panels have openings for 0.5 mm thick silicon windows for incoming and outgoing neutron beam to pass through.
The base structure of the system is a cut-to-fit optical breadboard that has 1 /4 in.-20 threaded holes in a 1 in.spacing pattern across the surface of the breadboard for easy installation and adjustment of the optical layout.The system implements a one-layer optical layout as shown in Figure 5.A 200 W fiber-coupled laser with FWHM = 0.35 nm is used to polarize the 3 He cell.The laser beam is split into two linearly polarized beams by the polarizing beam splitter.Two 3 in.diameter liquid crystal retarders serve as quarter wave plates and circularly polarize the two laser beams before they are directed to the 3 He cell.These retarders have a 500 W/cm 2 laser damage threshold and a reflectance less than 0.5%, 12 therefore they are well suited for our application.The cell named "Hokie" is fabricated in house using GE180 glass and it has an outer diameter of 7 cm and an overall length of 8 cm. 13 The cell is filled with a mixture of 2.3 bars of 3 He gas, 0.12 bar of N 2 gas, and traces of rubidium and potassium.It should be noted that the cell was not specifically designed and pressure optimized for the wavelength band 2-4.5 Å, but was the best choice of available cells.New cells that are optimized for a broader wavelength band of 2-8 Å are currently being fabricated.A µ-metal enclosed solenoid provides a 15 G magnetic field at the cell position, details about the solenoid can be found in our previous paper. 6he optical layout is shown in Figure 5.The stray light intensity in the system is monitored using a photodiode.This additional photodiode is interlocked with the laser.If the amplitude of the signal from this photodiode was to drop below a preset level due to breakage of the laser fiber, the laser will be shut off automatically to prevent the fiber from overheating and to avoid a safety hazard.Such a feature is essential to meet the safety regulations in user facilities.
The magnitude of the 3 He polarization is monitored in situ using a nuclear magnetic resonance (NMR) technique called free induction decay (FID) which provides a voltage readout that is proportional to the 3 He polarization.In addition, the photodiode (different from stray light photodiode) shown in Figure 5 provides the electron paramagnetic resonance (EPR) capability which measures the absolute value of 3 He polarization.Finally, to measure all four neutron spin cross sections during experiments with polarization analysis, the 3 He polarization needs to alternate between two states, i.e., + and ☞, in a timely manner.We have installed the adiabatic fast passage (AFP) mechanism that flips 3 He spins via a wave packet through electronics; [14][15][16]  The helicity of the circular polarization of the laser is controlled by adjusting the peak voltage of a square waveform applied to the liquid crystal retarders.The voltage value must also be reversed accordingly to continue to optically pump the cell in the new polarization direction after each AFP flip.The data acquisition control system (DAS) simultaneously sets pre-established new values for the liquid crystal peak voltages according to the desired final spin state as the analog I/O module is commanded to generate the RF voltage waveform for the AFP.This sequence is synchronized and automated from the data acquisition integrated Python scripting engine (PyDAS 17 ), according to the desired spin state changes scheduled by the user through its associated graphical user interface (GUI).During the three-week run, the liquid crystal retarders maintained stability and performed as desired.
The entire system weighs approximately 110 kg and is mounted onto a stainless steel plate.The system can be lifted out of the neutron beam or lowered down in the beam (see Figure 6) by using a motor driven linear slide.The dimension of the system is 69 cm in length, 59 cm in width, and 54 cm in height.
FIG. 6.The overview of the mounted system.The system is mounted on a motorized linear vertical slide and can be reproducibly moved down in the neutron beam and up out of neutron beam without being switched off.

IV. COMMISSIONING AND PARAMETERS
To investigate the performance of the 3 He system, we conducted a detailed study and series of measurements of the system in the experimental conditions on the Magnetism Reflectometer and commissioned it for operation in experiments.
First, the transmission measurements of the unpolarized neutron beam were conducted to derive the 3 He polarization and the cell thickness; the polarized neutron transmission measurements were used to measure the flipping ratios.
The unpolarized neutron transmission is described by the following two equations: T n (λ) = T 0 cosh(nσ 0 lλP He ), (7)   where T 0 is the transmission through unpolarized 3 He and T n the transmission through polarized 3 He.The first fit of T n T 0 (Figure 7) gives the product of nσ 0 lP He to be nσ 0 lP He = 1.098 ± 0.001 Å −1 .
The transmission through the cell glass windows and materials that are in the beam was measured previously to be T e = 0.82. 6rom the fit to T 0 T e (Figure 8), the cell thickness can be readily obtained as follows: Combining with the first fit result, we obtain the 3 He polarization to be (84 ± 1)%.The fits were performed using a standard least square refinement on the data up to 7 Å and using the covariance matrix to estimate the errors on the resulting parameters.The upper plot in Figures 7 and 8 shows the difference between model and data divided by the individual standard deviations.We attribute the rather large X 2 /degrees of freedom (DoF) values to systematic deviations of data points below ∼3.7 Å caused by the attenuation in the windows below a Bragg-edge.Fits performed on a limited data range above this wavelength yield X 2 /DoF close to 1 with no changes to 3 He polarization value but increased error bars.We have also confirmed the stability of our result by performing logarithmic fits to the data and by also fitting T e , which all resulted in 3 He polarization values between 82% and 85%.EPR measurement was carried out to confirm and crosscheck the 3 He polarization.The EPR splitting between the spin up and the spin down states was measured to be 26 KHz.Calculation 18,19 based on the pressure of the 3 He cell indicates a (85 ± 4)% 3 He polarization which is in good agreement with the polarization derived from the neutron experiment.
The neutron experiments at the MR beamline were performed with a neutron polarizer using a single reflection polarizing supermirror with m = 4 from SwissNeutronics 20 and an adiabatic radio-frequency gradient resonance spin flipper developed at SNS 21 to alter the polarization direction of the incoming neutron beam.To measure the flipping ratio, the polarized neutron transmission measurements were performed with the spin flipper turned on and off. Figure 9(a) shows the measured polarized neutron transmission with the neutron polarization parallel to the 3 He polarization, and calculated transmission using fitted parameters (cell thickness and polarization) obtained above and from the following equation: T e exp(−nσ 0 lλ) exp(nσ 0 lλP He ).
The polarization obtained from the measured flipping ratio is a convolution of the polarizer efficiency, the 3 He analyzing power, and the spin transport efficiency (Figure 9(b)).

V. EXPERIMENTAL TEST RESULTS
The 3 He system is installed at the Magnetism Reflectometer at the Spallation Neutron Source of Oak Ridge National Laboratory 22 as the neutron polarization analyzer for polarized neutron reflectometry (PNR) experiments.PNR employs the polarization analysis method to unravel the magnetization arrangement in thin films and multilayers.][25][26] Using PNR, the profile of magnetization vector distribution can be measured with an accuracy of a few kA/m with depth resolution ∼1 nm.][27] It was successfully applied to unravel the spiral magnetization vector distribution in exchange spring bilayers, 28 the domain structure in magnetic exchange-coupled multilayers 29 or even more complicated technologically relevant systems of laterally stripe-patterned exchange-biased films. 30uantum-mechanical effects of Larmor pseudo-precession of neutron polarization vector at reflection can also be observed. 31o discriminate the magnetic and non-magnetic offspecular scattering, polarization analysis of the scattered neutrons is needed for the divergent neutron beam across a broad range of wavelengths with a high level of analyzing power and good transmission.Our in situ 3 He analyzer is ideal for this purpose.
To examine the properties of the 3 He assembly, we performed a test experiment on a multilayer structure grown with molecular beam epitaxy on a sapphire substrate.The sample is exchange coupled single crystalline [ 57 Fe/Cr] × 12 superlattice in multi-domain state with four-fold symmetry. 32The thickness of the chromium layer of ≈9 Å provides antiferromagnetic (AF) orientation of magnetization in the alternating Fe layers.The in-plane external magnetic field was applied along one of the axes of easy magnetization and induces a spin-flop phase in the system with layer magnetization turned perpendicular to field.The sample was studied in detail before 32 and was chosen here for the test due to its rich features in spin-flip offspecular scattering.The PNR experiment was performed in an external magnetic field of 30 mT applied in plane of the sample after saturation in a field of 1 T. Using the polarized neutron beam with polarization 98.5% 33 and a wavelength band from 2.6 to 8.6 Å, a resonance spin-flipper in front of the sample and flipping the 3 He polarization as described in Section III, we measured the four "+ +," "☞ +," "+ ☞", and "☞ ☞" neutron spin cross sections.First/second signs designate the polarization of the neutron spins before/after the sample, and "+" and "☞" designate the neutron polarization with respect to the direction of the external field at the sample position.
The experiment is performed with two wavelength bands of 2.6 Å-5.6 Å and 3 Å-8 Å and several values of the incident angles to cover a broad band of momentum transfer Q. Figure 10 depicts the experimental two-dimensional intensity maps for "+ ☞" and "☞ ☞" neutron cross sections shown as functions of the perpendicular to the sample surface component of momentum transfer Q z and p i ☞ p f .The p i and p f are the perpendicular components of the incident k i and scattered k f neutron wavevectors (k i/f = 2πsin(θ i/f )/λ), as shown in the scheme at the right hand side of Figure 10 (see Ref. 34 for detailed explanation).The experiment confirmed the high quality of polarization analysis provided by the new 3 He assembly.
In addition, to test the stability of the 3 He analyzer, the system was left operating continuously for about 3 weeks.During this time, the 3 He polarization remained stable as measured with NMR.
The 3 He analyzer opens up new capabilities and enables us to perform experiments on the magnetic multilayers with noncollinear magnetization vector distribution in helical magnetic structures in Dy/Y superlattice, magnetic vector distribution in novel magnetic inherent nanolaminate Mn 2 GaC films, magnetization depth profile in continuous and laterally structured FeRh films.

VI. DISCUSSION
In this paper, we present the results on the redesign and manufacturing of the new in situ 3 He analyzer for the Magnetism Reflectometer.The 3 He polarization improved from 76% to 84%, as a direct result from better laser alignment and the use of a new fiber delivered, narrower spectrum width laser source.6][37] In the current setup, we made the change such that all optics are on the same layer which has greatly simplified the laser alignment.In addition, we used a 200 W fiber coupled laser (FWHM ∼ 0.35 nm) in the new system which reduced the number of laser shaping lenses and offered a more uniform power distribution across the laser beam.
Our 3 He system can accommodate cells up to 11 cm in diameter as compared to the current cell diameter of 6 cm.The use of a bigger cell will increase the solid angle coverage.Also, by shortening the distance between the sample stage and the 3 He system, the solid angle can further be enlarged.The fabrication of bigger cells and the new system design are both in development but outside the scope of this paper.New updates will be shown in future works.

VII. CONCLUSION
The advanced high performance in situ polarized 3 He system for the time-of-flight Magnetism Reflectometer at the SNS is developed.The 3 He polarization obtained with the new-generation optical pumping system establishes a new record value of (84 ± 1)%.To our knowledge, this is the highest polarization ever reached in in situ systems worldwide.The high 3 He polarization is achieved by using the new fiber delivered laser source and through an optimization of the laser alignment.The entire system is remotely controlled and monitored from the instrument control room including the NMR and EPR measurements.The increased 3 He polarization resulted in further enhancement of the transmission of the 3 He neutron polarization analyzer.The polarization stability has proved its reliability for several weeks of operation without interruption.

FIG. 1 .
FIG.1.The cell opacity as a function of the 3 He polarization.

FIG. 2 .
FIG. 2. The average figure-of-merit as a function of the neutron wavelength.
FIG. 4. (a)The scheme of the outer frame and the heater box; (b) the assembled system with laser safety panels installed.All the panels except the front main access panel use Torx screws to limit access; the positions marked with stars on the front panel have limit switches behind them.

FIG. 7 .FIG. 8 .
FIG.7.Ratio of unpolarized neutron transmission through the polarized (T n ) and unpolarized (T 0 )3 He cell as a function of the neutron wavelength (bottom) and residuals used for refinement (top, X 2 /DoF = 4.5).

FIG. 9 .
FIG. 9. (a)The polarized neutron transmission with the neutron polarization parallel to the3 He polarization as a function of neutron wavelength, the green solid line is the calculated transmission function using Eq.(9) with fitted parameters nσ 0 l = 1.311 and P He = 0.84.(b) Neutron polarization as a function of neutron wavelength.

FIG. 10 .
FIG. 10.Polarized neutron reflectometry (PNR) experiment.Left figure shows a schematic view of the PNR experiment with polarization analysis of specularly reflected k f (black arrow) and off-specular scattered k f (red arrow) intensity from a multilayer sample with magnetic domains shown with alternating burgundy and fuchsia colors.Neutron polarization appears as dark and light blue colors arrows, designating the direction of neutron polarization along or opposite to the direction of the external magnetic field H applied in plane of the film (shown with yellow color).Right figures display experimental two-dimensional (2D) maps of the intensity scattered from the Fe/Cr multilayer as a function of p i ☞ p f and Q z = p i + p f , with p i and p f the perpendicular to the sample surface components of the incoming k i and the outgoing k f wave vectors, respectively (see schematic in the left panel).The logarithmic intensity color scale is shown on the side.The incoming neutrons are in + (top figure) and ☞ (bottom figure) states, correspondingly, the reflected and scattered neutrons are in ☞ state for both top and bottom figures.