Reflective, polarizing, and magnetically soft amorphous neutron optics with 11B-enriched B4C

The utilization of polarized neutrons is of great importance in scientific disciplines spanning materials science, physics, biology, and chemistry. However, state-of-the-art multilayer polarizing neutron optics have limitations, particularly low specular reflectivity and polarization at higher scattering vectors/angles, and the requirement of high external magnetic fields to saturate the polarizer magnetization. Here, we show that, by incorporating 11B4C into Fe/Si multilayers, amorphization and smooth interfaces can be achieved, yielding higher neutron reflectivity, less diffuse scattering, and higher polarization. Magnetic coercivity is eliminated, and magnetic saturation can be reached at low external fields (>2 militesla). This approach offers prospects for substantial improvement in polarizing neutron optics with nonintrusive positioning of the polarizer, enhanced flux, increased data accuracy, and further polarizing/analyzing methods at neutron scattering facilities.


Supporting Information Section S1. Determination of 11 B4C content and amorphization conditions
To investigate the effects of 11 B4C on Fe-layer amorphization, roughness and formation of iron silicide, X-ray diffraction (XRD) analysis was performed of samples synthesized with different concentrations of 11 B4C.These samples were grown in a different sputtering chamber, see section S7.The crystalline nature of Fe/Si multilayers for 0 vol.%,2.5 vol.% and 5 vol.% is indicated by the presence of Fe (110) and/or Fe3Si (220) peaks that combine to form a broad diffraction peak around 45° 2.The width of this peak is a result of the presence of grains with varying sizes. 714 vol.% on the other hand was enough to yield X-ray amorphous multilayers, free from Fe or Fesilicide crystallites.It has been established that amorphous ferromagnets exhibit substantially lower coercivity than their crystalline counterparts.In order to verify this statement, the magnetic properties of an Fe/Si sample and an Fe/Si + 11 B4C sample with 14 vol.% 11B4C within the Fe and Si layers were studied using vibrating sample magnetometry (VSM).The results, depicted in Fig. S1B, reveal that the inclusion of 14 vol.% of 11 B4C indeed eliminated the coercivity, and resulted in a difference of more than an order of magnitude in the external magnetic field required to saturate the magnetization.To saturate the Fe/Si at least 10 mT were needed while Fe/Si + 11 B4C barely needed 1 mT.It Is important to note that the 14 vol.%lower Fe content in the 11 B4C incorporated sample leads to a decrease in the total magnetization, which can be observed in the amplitude drop in the 11 B4C incorporated sample.The slightly lower magnetization affects the m-SLD by decreasing it with the same ratio 8 but could, through our concept, easily be compensated for by increasing the non-magnetic layer SLD to achieve optimal polarization (Fig. 1B) which was then also done as seen in Fig. 4G.
The effect on the reflective properties for various concentrations of 11 B4C within the Fe/Si multilayers can be seen in Fig. S1C where the Bragg peak intensity increases with the increasing concentration of 11 B4C within the multilayer.The intention was to try to keep the same bilayer thickness for comparison, so the higher the 11 B4C content the lower the Fe and Si content.Although the nominal bilayer thickness was 25 Å, with increasing 11 B4C concentration the bilayer decreased in thickness, which is evident from the shift of the Bragg peak to higher scattering angles.However, this further solidifies the fact that the interface width decreased with higher amounts of 11 B4C since the Bragg peak increased even though the intensity should decrease due to the Bragg peak appearing at higher scattering angles, according to Porod's law.Further, the addition of 11 B4C in both layers decreases the X-ray SLD contrast, hence should in theory decrease the intensity of the Bragg peak with increasing concentrations of 11 B4C.However, apparent in Fig S1C , the Bragg peak still increased in intensity with higher concentrations, presenting evidence that the interface width decreased.However, too high concentrations of 11 B4C could reduce the neutron polarization capabilities due to a decrease in magnetization.Thus 14 vol.%within the Fe layers was chosen to maintain sufficient magnetization while still being magnetically soft for the continuation of our study.Observing Fig S1D, the incorporation of 11 B4C resulted in more intense Bragg peaks and higher orders of Bragg peaks for all measured bilayer thicknesses, supporting our idea of decreased interface width, regardless of bilayer thicknesses.The presence of Bragg peaks at higher Q-values enables the possibility of reflectivity at higher Q-regions.The figure also illustrates a Bragg peak for a bilayer thickness of 15 Å when 11 B4C is present, whereas no Bragg peak is observed without 11 B4C.GenX3 21 was used to fit the Fe/Si and Fe/Si + 11 B4C samples with 25 Å nominal bilayer thickness, as shown in Fig. S1E.The analysis suggests that the Fe/Si sample contains 26 Å of heavily mixed Fe and Si atoms throughout the entire sample which suggests iron-silicides throughout the entire multilayer, however with approximately 8 Å Si rich areas resulting in the Bragg peak that we see in Fig. S1E.The addition of 11 B4C prevented the formation of iron-silicide, resulting in a more clearly separated magnetic and non-magnetic layer, a decreased interface width, and increased reflectivity, along with an additional order Bragg peak.

Section S2. XPS and chemical bonding
Both Fe 2p and Si 2p peaks shift towards higher binding energy for the 11 B4C containing sample as compared to the Fe/Si sample.The shift is 0.65 eV and 0.45 eV, respectively.This can be explained by the formation of Fe-B, Fe-C, Si-B, and Si-C bonds.As both B and C are more electronegative than Fe and Si, the valence charge density on Fe and Si atoms is expected to decrease upon bonding.Thus, the hypothesis that the Fe-B bonds are what causes the amorphization is then confirmed.

Section S3. TEM and HAADF-STEM with EELS
From Fig. S3 it seems like the Fe/Si + 11 B4C multilayer does not have an even distribution of B throughout the multilayer or within each layer.There seems to be a B-poor region within the Si layers.Combined with the results from the XPS, in Fig. S2, it is suggested that due to Si-C bonds within the Si layer the B would have a smaller probability to be deposited in Si rich regions.Section S4.GIWAXS Fig. S4A and Fig. S4B shows the grazing-incidence wide angle X-ray scattering (GIWAXS) of Fe/Si and Fe/Si + 11 B4C multilayers, where Fig. S4A shows multilayers with a bilayer thickness of 100 Å and N = 10 and Fig. S4B multilayers with a bilayer thickness of 25 Å and N = 20.The GIWAXS data was taken simultaneously as the GISAXS measurements at the MiNaXS/P03 beamline using a LAMBDA 9M detector (X-Spectrum, pixel size = 55 µm) at a sample-to-detector distance of 284 mm.The background subtracted data is plotted using a temperature scale for intensity.Grazing-incidence small-angle X-ray scattering (GIWAXS), Fig. S4A and Fig. S4B, shows the crystallinity of Fe/Si multilayers with various concentrations of 11 B4C.The quarter circles and ordered intensities stem from crystallinity within the samples known as Debye-Scherrer rings.The intensity in the Fe/Si sample is concentrated on specific spots on the quarter circle, indicating a more defined ordering compared to a sample where the intensity is evenly distributed over the quarter circle.The broadness of the ring is related to the size of the crystallites.On the other hand, the GIWAXS pattern of the Fe/Si + 11 B4C sample shows a broad and featureless subtle ring, indicating a lack of well-defined crystallographic planes in the material.However, it is interesting to note that the Fe/Si + 11 B4C sample still retains this hint of the quarter circle seen in the Fe/Si sample.By adjusting the colormap intensity, it was possible to investigate a series of Fe/Si multilayers with varying concentrations of 11 B4C to examine the transition from crystalline to amorphous in detail, as seen in Fig. S4(b).The intensity of the quarter circle shows a clear trend of going from well-defined and intense crystallinity to non-existent with increasing 11 B4C concentration up to 40 vol.%.However, traces of crystallinity were detected up to 20 vol.% when compared to XRD results.This indicates that GIWAXS, with its higher sensitivity to crystallinity, was able to detect even small traces of crystallinity for higher 11 B4C concentrations.Overall, the results demonstrate the effectiveness of GIWAXS in identifying the presence of crystallinity in multilayer samples with varying concentrations of 11 B4C.
Section S5.Simulations of thicker bilayers using the extracted fitted parameters from Fig.

2.
Fig. S5 shows the simulated polarizing neutron reflectivity using the fitted parameters from Fig 2B .Fig. S5A represents the Fe/Si multilayer while Fig. S5B displays the Fe/Si + 11 B4C.It should be noted that the polarization results stated in the main text is related to the multilayers with the specified bilayer thicknesses and number of bilayers.For thicker bilayers, however, e.g.200 Å, the polarization values are here predicted to not be substantially increased unless finer SLD tuning is performed.
Using the fitted parameters obtained from Fig. 2B, thicker bilayers were simulated, demonstrating that the polarization typically approaches 100%.Our study aimed to reflect and polarize at scattering angles/vectors that are not yet utilized or achievable by state-of-the-art neutron optics.Incorporating 11 B4C into Fe/Si multilayers has been simulated to increase the polarization also for thicker bilayers as well although not as substantially as for thin bilayers where SLD matching is critical.The increase in polarization for a bilayer thickness of 200 Å was only 1%.Furthermore, it is theoretically possible to finetune the concentration of 11 B4C to achieve even higher polarization, but this requires smaller increments of 11 B4C to find the optimal amount for polarization enhancement.

Section S6. Elastic Recoil Detection Analysis (ERDA)
If the vol.% is a less desired measure, atomic percentages for all the samples obtained from elastic recoil detection analysis (ERDA) are presented in the tables below.Due to the depth resolution of ERDA being several nanometers and the samples of interest having layer thicknesses as small as a nanometer the following method was used to obtain the atomic percentages.All the samples measured with ERDA have the same amount of 11 B4C in the Fe layer but various amounts in the Si layer.By measuring the amount of 11 B4C in the whole multilayer of the Fe/Si + 11 B4C (13.9 vol.%) sample and a sample with twice the amount of 11 B4C in the Si layer but the same amount of 11 B4C in the Fe layer the amount atomic pecentage of 11 B4C within the Fe layer could then be calculated and using extrapolation also what percentages of 11 B4C in Si the other samples should have.ERDA measurements on those other samples confirmed that the extrapolation were accurate with less than a percent's margin.Important to note however is that the atomic percentages within each layer is based on the volumetric percentages which in turn may not be very accurate as seen through XRR fits and TEM, due to the layer thicknesses differing from the nominal values.ERDA measures the amount of each atom and since 11 B and C are not stoichiometrically 4:1 when deposited in films the atomic percentages show the total amount of 11 B atoms + C atoms.Another reason to count 11 B and C together is due to the fact that 11 B and C intensities appear too overlapped in the ToF-ERDA spectra to be able to separate them well.Although an approximate separation was made and can be seen in Table 3. Tables 1, 2 and 3 shows the atomic compositions from ERDA of all multilayers.From Table 1 together with Fig. S1A we can conclude that by replacing every fifth Fe atom with a 11 B (or C) atom the Fe layer becomes X-ray amorphous and magnetically soft.From Table 2 along with Fig. 4C we obtain the atomic percentages of 11 B + C needed to achieve optimal polarization for 25 Å bilayer thickness and a thickness ratio of 0.5.However, since the thickness ratio turned out not to be 0.5 the exact percentages may not be very accurate, although the relation between the samples are.From Table 3, it can be concluded that the actual composition of the incorporated 11 B4C is closer to 11 B~5C.Section S7.Experimental details for deposition system for samples from S1, S3 and Fig. S4B Fe/Si and Fe/Si + 11 B4C multilayer thin films were deposited using ion-assisted magnetron sputter deposition in a high vacuum system with a background pressure of about 5.6•10 -5 Pa (4.2•10 -7 Torr).The multilayers were deposited onto 001-oriented single crystalline Si substrates, 10×10×1 mm 3 in size, with a native oxide.During deposition the substrate temperature was maintained at an ambient temperature (293 K), and to improve the thickness uniformity, the substrate was spinning at 8 rpm.The substrate table was electrically isolated, enabling a substrate bias voltage to be applied in order to attract sputter gas ions from the plasma.The main difference growing these samples compared to the other sample in this article was by using a modulated ion assistance regime during the deposition. 27Ion-assisted deposition was employed by attracting Ar-ions from the sputter plasma through a negative substrate bias of -30 V.A magnetic field, colinear with the substrate normal, was used to condense the plasma towards the substrate, thereby increasing the Ar-ion flux at the substrate.The sputtering was modulated by alternating between 0 V substrate bias for the first approximately 3 Å to then employ the -30 V substrate bias for the rest of the layer.The modulated ion-assistance scheme is for the decrease in intermixing.The sputtering targets used were Fe (99.95% purity, 75 mm diameter), 11 B4C (99.8% chemical purity, isotopic purity >90%, 50 mm diameter), and Si (99.95% purity, 75 mm diameter).The magnetrons were continuously running during deposition and the material fluxes were controlled using computercontrolled shutters placed in front of the magnetrons for each target material.This allowed for deposition of multilayers from the separate target materials, as well as the possibility of alloying two target materials to achieve a desired composition through co-sputtering.When depositing Fe/Si + 11 B4C, each bilayer consisted of co-sputtered 11 B4C with Fe, followed by the deposition of 11 B4C with Si.The deposition rates of both Fe + 11 B4C and Si + 11 B4C were nearly equal, approximately 0.5 Å/s.When preparing samples with different ratios of Si to 11 B4C, only the deposition rate of Si was adjusted by tuning its target power.

Fig. S3 .
Fig. S3.Microscopy on the B and Fe out-of-plane spread.(A) Section of overview diffraction contrast TEM micrograph of Fe/Si + 11 B4C multilayer with 14 vol.% of 11 B4C from Fig. 4(D to F) and (B) corresponding scanning HAADF-STEM image and overlayed 150 x 150 Å 2 electron energy loss spectroscopy (EELS) map of Fe and B. From elemental analysis obtained using XPS, TEM, STEM, and EELS a schematic illustration of the elemental distribution in the multilayer stack is shown in the image.

Fig. S4B .
Fig. S4B.Transition from crystalline to amorphous structure with increasing 11 B4C concentration.GIWAXS patterns of Fe/Si multilayers with varying concentrations of 11 B4C, with a fixed bilayer thickness of 25 Å and N = 20.Samples grown in a different chamber, see S7.