Structure and magnetism of the triangular lattice material YbBO3

YbBO3 is a member of the orthoborate family of materials which contains a triangular arrangement of Yb3+ ions. Here we study the physical properties of YbBO3 with neutron diffraction, inelastic neutron scattering, specific heat, and ac susceptibility measurements. The neutron diffraction measurements confirm that our samples of YbBO3 crystallize in the monoclinic space group C2/c (#15) which contains two crystallographically distinct Yb3+ sites decorating a slightly distorted triangular lattice. Heat capacity and ac susceptibility measurements indicate a potential transition to magnetic order at 0.4 K. In agreement with these observations, neutron diffraction measurements at 0.044 K observe magnetic Bragg peaks which can be indexed by a propagation vector of (0 0 1). Although determining a unique spin configuration corresponding to the observed magnetic Bragg peaks is not possible, model refinements indicate that the ordered moments are likely in the range of 0.4–0.9  μB and, notably, require moments on both Yb sites. In addition to the magnetic Bragg peaks, diffuse scattering at low Q is observed indicating that the transition does not correspond to complete long range magnetic order. The two-site picture for YbBO3 is further evidenced by the number of crystal field excitations observed by inelastic neutron scattering measurements. Together these results show that YbBO3 is a two-site triangular lattice material with signatures of long-range order as well as shorter ranged spin correlations.


Introduction
Triangular lattice antiferromagnets are of enduring interest as prototypes for quantum magnetism [1][2][3][4] as well as for their potential to host spin liquid phases [5][6][7][8][9][10]. At the fundamental level, the key feature of the triangular lattice is the strong geometric frustration of the nearest neighbor spin-spin interactions. Nevertheless, if the spin-spin interactions are isotropic and strictly nearest neighbor, the ground state has conventional magnetic order with a simple 120 • structure [11][12][13][14]. Hence, more exotic phases (e.g. spin liquids) can only emerge on the triangular lattice in the presence of further neighbor and/or anisotropic interactions between spins [15][16][17][18].
Traditional investigations of triangular lattice antiferromagnets focused on materials where the magnetism is derived from a 3d element [20][21][22][23][24][25][26]. More recently, however, rare earth materials have also gained attention as several Yb-based compounds have been put forward as spin liquid candidates; for example, YbMgGaO 4 [27][28][29][30][31][32], NaYbO 2 [33][34][35], NaYbS 2 [36,37], and MYbSe 2 [38][39][40][41] with M = Na, K, Cs. Still, finding an ideal triangular lattice material can be challenging since the ground state is very sensitive to further neighbor and interlayer interactions [41] and its identification may be clouded by various types of defects and disorder [42][43][44]. The rare earth orthoborates [45] in general and YbBO 3 in particular offer a potential avenue to investigate such issues. YbBO 3 has been previously reported to crystallize in several different structures [45][46][47][48]. The original report was a rhombohedral crystal structure [46], but a recent comprehensive study of the series of RBO 3 (R = Gd, Tb, Dy, Ho, Er, Yb) by Mukherjee et al [45] has shown using powder x-ray diffraction that the materials crystallize in a monoclinic structure (space group #15). The lower symmetry crystal structure has important implications for the magnetic properties. More specifically, there is no longer an ideal triangular lattice in this monoclinic structure and there are two distinct Yb sites with different local symmetries and multiplicities. Additionally, measurements of the magnetic susceptibility down to 2 K and specific heat measurements down to 0.4 K did not find evidence for long range magnetic order, though there was a strong upturn in the specific heat down to 0.4 K [45]. Most recently, specific heat and µSR measurements did not observe a sharp peak in the heat capacity above 0.4 K and the characteristic features of long range magnetic order expected in a µSR measurement were not observed down to 20 mK [49].
In this article, we study the physical properties of the triangular lattice material YbBO 3 through thermodynamic, neutron diffraction, and inelastic neutron scattering measurements of polycrystalline samples. We use neutron diffraction to confirm the crystal structure. As in [45], we find that YbBO 3 has a monoclinic crystal structure with two Yb sites with small deviations from an ideal triangular lattice (see figure 1). Both heat capacity and ac susceptibility indicate a transition to magnetic order at T = 0.4 K. Low temperature neutron diffraction measurements find evidence for spin ordering at 0.044 K with magnetic moments on both Yb sites in the range of 0.4-0.9 µ B .  (see table 1). (b) A single triangular lattice layer with the Yb1 and Yb2 sites indicated. The triangular lattice is not an ideal equilateral triangular lattice, rather there are six distinct lengths for the sides of the triangles ranging from 3.684 to 3.816 Å. The shortest Yb-Yb distance along the c-axis is 4.373 Å. The crystal structure was drawn using Vesta [19].
The low temperature neutron diffraction measurements additionally reveal significant diffuse scattering indicating that the transition does not correspond to complete long range magnetic order. Inelastic neutron scattering measurements indicate the presence of four crystal field excitations above the ground state, providing further support for the description of the crystal structure with two unique Yb sites.

Experimental details
Polycrystalline samples of YbBO 3 were prepared by traditional ceramic processing methods. Initially, B 2 O 3 and dried Yb 2 O 3 were ground together, pressed into a pellet and fired in air to 900 • C. The material was reground, pressed into a rod using a balloon and hydrostatic press, and then fired at 1000 • C for 12 h. After this, some Yb 2 O 3 could still be observed by xray diffraction, and thus additional B 2 O 3 was added with additional firings to 1000 • C until the laboratory x-ray diffraction pattern appeared clean (see figure A2). While the laboratory diffraction may not detect B 2 O 3 easily, it is a preferred impurity over magnetic Yb 2 O 3 . To minimize neutron absorption, the samples were made using isotopically enriched B (99.62% 11 B, supplied by Sigma-Aldrich). Samples were stored in a glovebox to avoid possible absorption of moisture. No evidence of an impurity phase was observed in the neutron diffraction measurements presented here or in final laboratory x-ray measurements. Magnetization as a function of temperature is presented in figure A1 as an additional check of sample purity; specifically, no evidence for Yb 2 O 3 , which has an antiferromagnetic transition at 2.3 K [50], is detected in the magnetization or specific heat measurements.
Heat capacity measurements at constant pressure (C p ) down to 0.066 K were performed in a dilution refrigerator insert in a Quantum Design Dynacool. The measurements were made on a small pellet that was formed by pressing the enriched YbBO 3 sample with an equal mass of high purity Ag powder. The Ag powder was added to bolster the mechanical integrity of the pellet, increase thermal transport, and to promote thermal equilibrium throughout the sample. In-house data for the specific heat of silver were utilized to correct the measured specific heat for the contribution of silver, which becomes negligible at low temperatures; proper addenda correction for the Apiezon N-grease were also performed. As a check of batch consistency, samples were measured from two separate batches and the results were consistent. AC susceptibility measurements were performed in the same dilution refrigerator using material from the same pellet as that made for specific heat, with a driving amplitude of 2 Oe and frequencies of 227 Hz and 756 Hz.
Neutron diffraction data at 10 and 300 K were collected using the POWGEN instrument at the Spallation Neutron Source (SNS) [51] located at Oak Ridge National Laboratory. Data were collected using a sample with a mass of 2.994 g. The sample was loaded in a vanadium sample can. The high resolution mode of the instrument with a wavelength band centered at 0.8 Å was used. Rietveld refinement of the structural model was performed with the GSAS-II software package [52].
Neutron powder diffraction measurements to study the low temperature magnetic scattering in YbBO 3 were performed using the HB2A powder diffractometer [53] at the High Flux Isotope Reactor (HFIR) located at Oak Ridge National Laboratory. A wavelength, λ = 2.406 Å, with premonochromator, presample and predetector collimator settings of openopen-12' was used for the measurements. The sample used included the same sample used in the POWGEN experiment, but with additional material from the same batch for a total sample mass of 3.802 g. The sample was loaded in a copper can with 1 atmosphere of helium to promote thermalization.
Inelastic neutron scattering measurements of the crystal field excitations were performed with the SEQUOIA spectrometer at the SNS. Data were collected with incident energies of E i = 300, 120, 60, and 11 meV. The high flux Fermi chopper was used for the highest incident energy, with the chopper spinning at 420 Hz, while the high resolution Fermi chopper was used for the latter incident energies with the chopper spinning at 600, 420, and 180 Hz respectively. For these chopper configurations, the energy resolution at the elastic line is calculated to be: FWHM= 20.7 meV for E i = 300 meV, FWHM= 3.6 meV for E i = 120 meV, FWHM= 1.6 meV for E i = 60 meV and FWHM= 0.24 meV for E i = 11 meV. The sample was the same one used for the HB2A experiment, but it was loaded in an aluminum sample can with ∼1 atmosphere of helium gas. A similar can also with ∼1 atmosphere of helium gas was measured under identical conditions using the SEQUOIA powder sample changer. The empty can subtraction was deemed necessary to remove the Al phonon scattering which peaks at approximately 20 and 30 meV in energy transfer. The empty can measurement was subtracted from all the inelastic neutron scattering measurements presented here unless otherwise noted.

High temperature neutron diffraction measurements
We start by examining the structural properties of YbBO 3 with neutron diffraction measurements. Neutron diffraction data were collected with the POWGEN instrument at 300 K and 10 K. Due to the similarity of the structure at 300 K and 10 K, here we focus discussion on the 300 K neutron diffraction data (see figure 2). For completeness, the results of the 10 K refinement are reported in appendix B (table B1 and figure B1).
Given the differing reports of the crystal structure of YbBO 3 (e.g. see [45][46][47][48]), we chose to reexamine three previously reported structures. The hexagonal structures P6 3 /m (single Yb site), P6 3 /mmc (single Yb site), and P6 3 /mcm (two Yb sites) were checked and did not adequately describe the neutron scattering data. Subsequently, we considered the monoclinic structure (space group C2/c with two Yb sites) reported by [45] as a starting point for additional model refinements. As shown in figure 2, this monoclinic model (see table 1) provides a good description of the neutron diffraction data. However, we note that as can be seen in figure 2, there are several peaks where the mononclinic model exceeds the measured intensity. At present the origin of this discrepancy is unclear, but may be related to preferred orientation or possibly the partial occupancies of the B and O sites discussed below.
With our neutron scattering data, we were able to refine the atomic positions and thermal parameters within the Table 1. Structural parameters from model refinement to POWGEN data at 300 K. Goodness of fit, Rw = 4.24 %. Parameters were refined for the monoclinic space group C2/c (#15) with a = 11.199(1) Å, b = 6.4740(1) Å, c = 9.4872(8) Å, α = 90 • , β = 112.771(2) • , γ = 90 • , unit cell volume = 634.21(2) Å 3 , and calculated density = 7.291 g cm −3 . One-σ values of the statistical error are are given as the value in parentheses in this table and in other values described in the manuscript. Parameters without errors were held fixed at the values given in the table. To obtain an estimate of fractional site occupancies, a distinct refinement protocol was adopted as described in the text. The resulting site occupancies are given in [  monoclinic structure. However, due to coupling between the thermal parameters and site occupancies we found that it was not possible to obtain a reliable refinement when both parameters were allowed to vary simultaneously. The following protocol was adopted to address this issue: thermal parameters of the same element were constrained to have the same value and the site occupancies were fixed to a value of one (full occupancy). Important refinement parameters are given in table 1. These results provides confirmation of the monoclinic structure, which contains two Yb sites with C 1 and C −1 symmetry and indicates that the individual triangular units are not equilateral. These deviations from ideal triangular geometry potentially relieves the frustration expected for a canonical triangular lattice and allows for spin configurations distinct from the 120 • expected for the canonical nearest neighbor Heisenberg triangular lattice [11][12][13][14]. As described below, this finding also has an impact on the observed crystal field excitations.
In order to obtain an estimate for deviations from the ideal stoichiometry, the following protocol was adopted: the site occupancies of the Yb atoms were fixed to 1 as individual refinements of these occupancies resulted in values within one σ of full occupancy. The site occupancies of the four oxygen atoms occupying general positions were constrained to be the same. The remaining site occupancies were allowed to vary independently. In addition, the thermal parameters and atomic positions were held fixed. The site occupancies from this refinement protocol are reported in italics in table 1. These results indicate that there are likely deficiencies of both B and O from the nominal composition.

Heat capacity and AC susceptibility measurements
Heat capacity and AC susceptibility measurements were performed to study the low temperature behavior of YbBO 3 . The specific heat data is shown in figure 3. At zero field the heat capacity exhibits a sharp peak at 0.4 K suggesting that a phase transition occurs at this temperature. In addition to the peak at 0.4 K, there is an upturn at lower temperatures that is especially visible when plotted as C p /T. The origin of this diverging C p /T is unclear. Integrating the C p /T data above 0.066 K shows that the expected value of Rln(2) = 5.76 J mol −1 K −1 is obtained by ≈0.8 K. The small upturn from 0.066 to 0.110 K contains approximately 0.2 J mol −1 K −1 and without this contribution the expected entropy of the doublet ground state is recovered by T = 1 K. A correction for any phonon background was not applied during this analysis, though the contribution is expected to be negligible due to the low temperature examined.
AC susceptibility measurements provide complementary information to the specific heat capacity measurements. The transition temperature of 0.4 K revealed by the heat capacity measurements is corroborated by the ac susceptibility measurements shown in figure 4. The in-phase contribution to the ac susceptibility (χ ′ ) has a peak at 0.4 K when a dc field is not applied, and application of a small dc field (1 kOe) broadens the peak and suppresses it to near 0.3 K. In the zero field data, there is an indication of a broad shoulder near 0.1 K, which may relate to the up-turn observed below this temperature in the specific heat data. A small out-of-phase component (χ ′ ′ ) emerges below ≈0.45 K. The out-of-phase contribution is typically negligible in antiferromagnetic materials as it commonly relates to losses from domain effects or electronic sources, and thus these results may indicate some canting of the moment develops upon cooling. The data in figure 4 were collected using a frequency of 756 Hz, and the peak in χ ′ occurred at the same temperature when a frequency of 227 Hz was utilized. Qualitatively similar behavior was observed for χ ′ ′ for these two different frequencies.

Low temperature neutron diffraction
Neutron diffraction measurements were performed at low temperatures to determine if the features in the heat capacity and AC susceptibility data correspond to the onset of magnetic order. Figure 5(a) shows the low Q part of the neutron powder diffraction results of YbBO 3 measured at T = 0.044(4) K and 0.700(4) K with the HB2A powder diffractometer at the HFIR. Scans of the diffractometer's scattering angle, 2θ, coverage at T = 0.044(4) K were collected on ∼2 h intervals for a total of ∼34 h. This experimental protocol allowed scans to be checked sequentially for proper thermalization. No significant differences between scans were observed, such that all scans, except the first 2 h scan, were summed and used for analysis. The data collected at T = 0.700(4) K (above the putative ordering temperature of 0.4 K observed in the thermodynamic measurements) were collected in 2 h scans for a total of 16 h. No significant difference between the scans was found, but like the 0.044 K data, the first 2 h scan was excluded as a precautionary measure. The remainder of the data were summed to improve counting statistics and the resulting data were used for analysis.
Carefully examining the neutron diffraction data in figure 5(a) reveals only small differences between the data acquired at the two temperatures. Given the long data collection times, this already suggests that any ordered moment must be small. Additionally, there is a noticeable difference between the data sets for Q < 0.5 Å −1 . This additional contribution at the lowest temperature measured is much sharper in Q than the Yb 3+ form factor indicating the buildup of significant spin-spin correlations. Subtracting the data set at 0.700 K from 0.044 K reveals additional detail. This difference is shown in figure 5(b). The additional scattering at the lowest Q is present to some degree until approximately 0.75 Å −1 . In the subtracted data, the baseline values are negative. This is consistent with the presence of uncorrelated paramagnetic scattering above the ordering temperature. Beyond this broad scattering at low Q, several magnetic Bragg peaks are revealed in the difference curve.
Given the small number of potential magnetic Bragg peaks and the large number of magnetic atoms per unit cell, uniquely determining the ordered spin configuration is not possible with the present data. However, some important information can still be extracted. As a starting point, we have attempted to find a propagation vector which explains the magnetic Bragg peaks. The commensurate propagation vector (001), i.e. a Q = 0 antiferromagnet, indexes the observed magnetic peaks. Assuming this propagation vector, the possible maximal magnetic subgroups were examined using the tools provided by the Bilbao Crystallographic Server [54]. This yields four potential magnetic space groups: C2 ′ /c ′ (#15.89), C2/c ′ (#15.88), C2 ′ /c (#15.87), and C2/c (#15.85). Of these, C2/c ′ and C2 ′ /c do not allow moments on the Yb2 site and do not produce intensity at the (001) peak position and thus can be discarded as possible solutions. Refinement of model parameters of the remaining two space groups were able to reproduce the scattering pattern, but for each group somewhat different magnetic structures were possible and it is furthermore not possible to distinguish between solutions in the C2 ′ /c ′ and C2/c space groups. Despite this ambiguity, the possible solutions have several similarities. Notably, all viable solutions require moments on both Yb sites with moment magnitudes in the range of 0.4-0.9 µ B . A more complete understanding of magnetic structure of YbBO 3 requires measurements of a single crystal and/or additional means of constraining model parameters.

Crystal field level splitting
Given that there are two distinct Yb sites per unit cell, it is interesting to investigate potential differences at the single ion level imparted by the crystal field level splitting of the ground state f -electron manifold of the Yb 3+ ions in YbBO 3 . As described above, the two Yb sites in YbBO 3 are rather low symmetry with only the identity (C 1 symmetry) and inversion operations (C −1 symmetry) allowed for the Yb1 and the Yb2 sites respectively. In principle, each Yb site will have unique crystal field level splitting. We address this issue through consideration of inelastic neutron scattering measurements of the crystal field excitation spectrum.
We start our discussion of the crystal field splitting in YbBO 3 by examining the inelastic neutron scattering measurements made with the SEQUOIA spectrometer at the SNS. Figures 6(a)-(d) shows the inelastic neutron scattering spectrum collected at 5 K for E i = 300, 120, 60 and 11 meV respectively. Figure 6(a) reveals a broad band of scattering centered near 70 meV which decreases in intensity with increasing Q. This is consistent with a crystal field excitation(s) which should show an intensity that decreases as a function of increasing Q according to the square of Yb 3+ magnetic form factor. The improved energy resolution provided by the 120 meV data in figure 6(b) shows that the broad line is actually composed of several distinct crystal field excitations. The data in figures 6(b) and (c) shows additional scattering intensity that increases with Q at several different bands of energy transfer. These are consistent with lattice excitations, i.e. peaks in the phonon density of states. Figure 6(d) indicates no appreciable magnetic scattering for T =5 K for energy transfers below 8 meV.
Additional detail concerning the crystal field data can be gleaned by examining cuts through the inelastic neutron scattering data as a function of temperature and energy transfer,hω, which are shown in figure 7. Figure 7(b) shows the scattering intensity as a function of energy transfer for different regions of integration in wave-vector transfer for the T = 5 K and E i = 300 meV measurement. There is a large peak centered at 70 meV that decreases in intensity with increasing wave-vector transfer. A lower value of incident energy, E i = 120 meV, as shown in figure 7(a), is able to resolve additional structure in this peak. There are several excitations of different origin evident in this data. Many of these, particularly at lower energy transfers exhibit an (a) Inelastic neutron scattering spectra of YbBO 3 measured at E i = 120 meV for several temperatures using the SEQUOIA spectrometer (see figure C1 for the full temperature dependent scattering patterns). Data are shown integrated between Q = 0 and Q = 5.5 Å −1 . Data have been background subtracted using the empty can measurement as described in the text. The lowest intensity data corresponds to the difference between the T = 5 K and the T = 200 K measurement as shown in the figure legend. The solid line through these data corresponds to four Gaussian peaks with no background as described in the text. The arrows at the top of the figure illustrate the refined peak positions based upon the four Gaussian fit. (b) Inelastic neutron scattering spectra of YbBO 3 measured at E i = 300 meV and T = 5 K using the SEQUOIA spectrometer. Data are shown integrated over three different ranges of wave-vector transfer as indicated in the figure legend. Data have been background subtracted using the empty can measurement as described in the text. increase in intensity as temperature is increased. This behavior is attributed to the thermal population of the phonon modes in this energy range. On the other hand, there are several modes at higher energy transfer that decrease in intensity with increasing temperature. These modes also have the Q-dependence expected for a crystal field excitation. This observation confirms their origin as crystal field transitions. To better isolate potential crystal field transitions, the data at T = 200 K were subtracted from data at T = 5 K, shown as a pink diamond data points in figure 7(a). From this difference we can identify four crystal field excitations. We fit the resulting lineshape to the sum of four Gaussian peaks (solid blue line in figure 7(a)) yielding crystal field energies of 55.7(1), 67.83(7), 76.35 (7), and 86.9(2) meV.
To understand the inelastic neutron scattering data, we first note that the ground state manifold of the Yb 3+ ion contains four Kramers-doublets in the absence of a field breaking time reversal symmetry. As one of the doublets must be the ground state, no more than three crystal field excitations should be present in an inelastic neutron scattering measurement. However, each of the two crystallographic sites in YbBO 3 can have different energies as determined by the local environment. The four crystal field excitations noted above are consistent with this argument.
To proceed we consider the CEF Hamiltonian. Specifically, the HB2A data confirmed the presence of two inequivalent Yb sites, one with with C 1 and one with C −1 symmetry, which means that the CEF Hamiltonian consists of more than 23 parameters per site [55]. Considering that Kramer's degeneracy will still be present, we are left with at most three measurable transitions for the lowest energy ground state J-multiplet per site. This small number of experimental observables in contrast to the large number of parameters in the Hamiltonian precludes a quantitative fit of the inelastic neutron scattering data. Hence, we examine a point charge model (PCM) to provide a qualitative basis to understand the observation of four crystal field excitations. From a PCM calculation we can predict the spectrum of the two sites in both LS-coupling and intermediate coupling as follows: LS-coupling and intermediate coupling are in reasonable agreement for the first three excited energy levels for each respective site (note the intermediate coupling scheme not only considers the ground state J-multiplet, but also the higher lying J-multiplets). These qualitative results infer the possibility that each Yb site gives rise to a series of distinct but overlapping CEF levels. Considering the LS-coupling calculation or the intermediate coupling for site 1 and site 2, and the closely spaced nature of some of the energy levels, the observation of four transitions between 20 and 100 meV energy transfer is reasonable, rather than only three for Yb (J = 7/2). Coupling neutron scattering data with Raman spectroscopy measurements could help shed additional light on the crystal field splitting and identify the correct eigenfunctions of the GS for the two Yb-sites.

Discussion
The specific heat capacity and magnetic susceptibility data indicate a transition to a magnetically ordered phase below 0.4 K. The recovery of an entropy of Rln(2) as determined from our heat capacity measurements further indicates that the majority of the Yb spins participate in this magnetically ordered phase. Additionally, the low temperature neutron diffraction data further shows the build up of a broad scattering signal at low-Q (below 0.75 Å −1 ) along with several magnetic Bragg peaks characterized by a propagation vector of (001). This latter observation is evidence that a substantial portion of the sample undergoes long range magnetic order.
Open questions remain concerning the relationship between the crystal structure and magnetic properties of YbBO 3 . The materials prepared and studied here, like those of [45], crystallize in the mononclinic space group C2/c. While this crystal structure retains triangular nets, the triangular nets are no longer ideal equilateral triangles with six distinct lengths ranging from 3.684 to 3.816 Å. This may allow for a departure from the 120 • order expected for an ideal Heisenberg model on the triangular lattice. On the other hand, recently, Somesh et al [49], have reported that their samples crystallize in the P6 3 /m space group which appears to be reflected in the physical properties of their samples compared to those measured here. For example, they observed a broad peak in their heat capacity measurements, and their muon spin resonance measurements do not find a signature of long range order. Instead, these measurements likely indicate dynamic spin correlations potentially due to nonmagnetic site disorder. In the samples studied here, the sharpness of the anomaly in the specific heat capacity, the lack of appreciable impurity phases, as well as the observation of only four crystal field excitations provides some evidence for a reasonably well ordered material. On the other hand, like [49], we also find evidence for nonideal stoichiometry in our samples and this is likely a factor in the observed diffuse signal at low Q in the low temperature neutron diffraction data. Future work which focuses on how different synthetic protocols realize different structures and magnetic behavior of YbBO 3 will be illuminating and may establish this as an important material for studying quantum magnetism.

Conclusion
In conclusion, we have studied the triangular lattice material YbBO 3 . Our work confirms the presence of two unique Yb sites with structural refinements to neutron diffraction data as well as measurements of the crystal field excitation spectrum. Specific heat capacity, ac susceptibility, and low temperature neutron diffraction reveal the onset of strong spin-spin correlations near 1 K with an anomaly in specific heat at 0.4 K suggestive of long-range order. Although we were unable to determine a unique magnetic structure from the data reported here, model refinements indicate ordered moments on both Yb sites in the range of 0.4-0.9 µ B . The low temperature neutron diffraction pattern also contains a diffuse component at low Q indicating that both long-range order as well as shorter ranged spin correlations occur in YbBO 3 .

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files). 2 for simplicity. The moment reached at 2 K and 65 kOe was assumed to be the maximum moment and the data collected in the 3 He probe were scaled to match this maximum due to small errors introduced by the nature of loading of the 3 He system; the maximum moment in the 3 He probe reached 1.99µ B /Yb at 0.4 K, which is a reasonable result.  Table B1. Structural parameters from model refinement to POWGEN data at 10 K. Goodness of fit, Rw = 4.5. Parameters were refined for the monoclinic space group C2/c (#15) with a = 11.1803(8) Å, b = 6.4631(1) Å, 9.4818(6) Å, α = 90 • , β = 112.75(2) • , γ = 90 • , unit cell volume = 631.86(1) Å 3 . Parameters without error bars were not refined.

Appendix B. POWGEN neutron diffraction data and structural refinement at 10 K
This appendix provides the details of refinements of the structural model of YbBO 3 against neutron diffraction data collected with POWGEN at 10 K.

Appendix C. Temperature dependent inelastic neutron scattering data
This appendix presents inelastic neutron scattering data with E i = 120 meV as a function of temperature. These data sets were used to produce the cuts shown in the main text in figure 7.