Unconventional Spin State Driven Spontaneous Magnetization in a Praseodymium Iron Antimonide

Consolidating a microscopic understanding of magnetic properties is crucial for a rational design of magnetic materials with tailored characteristics. The interplay of 3d and 4f magnetism in rare‐earth transition metal antimonides is an ideal platform to search for such complex behavior. Here the synthesis, crystal growth, structure, and complex magnetic properties are reported of the new compound Pr3Fe3Sb7 as studied by magnetization and electrical transport measurements in static and pulsed magnetic fields up to 56 T, powder neutron diffraction, and Mößbauer spectroscopy. On cooling without external magnetic field, Pr3Fe3Sb7 shows spontaneous magnetization, indicating a symmetry breaking without a compensating domain structure. The Fe substructure exhibits noncollinear ferromagnetic order below the Curie temperature TC ≈ 380 K. Two spin orientations exist, which approximately align along the Fe–Fe bond directions, one parallel to the ab plane and a second one with the moments canting away from the c axis. The Pr substructure orders below 40 K, leading to a spin‐reorientation transition (SRT) of the iron substructure. In low fields, the Fe and Pr magnetic moments order antiparallel to each other, which gives rise to a magnetization antiparallel to the external field. At 1.4 K, the magnetization approaches saturation above 40 T. The compound exhibits metallic resistivity along the c axis, with a small anomaly at the SRT.


Introduction
Intermetallic compounds based on rare-earth (RE) and transition metal (TM) elements have been investigated thermal hysteresis with a net magnetization antiparallel to the applied field. While diamagnetism due to superconductivity can be ruled out, the origin of this very unusual behavior was not further resolved.
Antiparallel magnetization has been reported in literature for numerous compounds with various underlying mechanisms. [22][23][24][25] Most of them share a strong temperature and field dependence of their magnetic properties. As a consequence, they possess a high potential for field-or temperature-induced magnetic switching in magnetic memory devices. [26,27] Besides, ferromagnetically ordered compensated systems have been investigated for their application in spin resolving devices. [28] Preparing phase pure samples in RE-Fe-Sb systems may be very challenging, as intricate phase equilibria between neighboring ternary compounds exist. Furthermore, the generally high melting points of the binaries complicate conventional growth from the melt, which demands the use of more sophisticated synthetic procedures. In this work, we report on the synthesis of high-quality crystals and powders via flux growth technique of the new compound Pr 3 Fe 3 Sb 7 , which shows equally interesting properties as the aforementioned Nd 3 Fe 3 Sb 7 . Single crystal X-ray diffraction confirms that it is isostructural to the Nd compound. We investigated the complex magnetic properties of single crystalline Pr 3 Fe 3 Sb 7 in static and pulsed magnetic fields and wide temperature ranges by means of powder neutron diffraction, Mößbauer spectroscopy, SQUID, and vibrating sample magnetometry as well as electrical transport measurements.

Results and Discussion
We synthesized Pr 3 Fe 3 Sb 7 crystals using an inert, molten Bi flux as reaction medium. Pr 3 Fe 3 Sb 7 specimens grow in a distinct needle-like shape of up to several mm length (Figure 1a). The separation of the ternary phase from the flux was achieved by hot centrifugation and wet chemical treatment. In the reported case, the molten metal Bi acts as the solvent for Pr and Sb and promotes the formation of the target compound, but is not incorporated in the structure. Compared to conventional sintering at elevated temperature, competing phases such as PrFeSb 3 or PrFe 0.6 Sb 2 are much better suppressed in the flux. Mechanochemical activation of the starting materials by ball milling did not significantly improve the result, which illustrates the thermodynamic limitations of a sub-solidus treatment. Furthermore, as the target compound decomposes peritectically, crystal growth from a melt of the elements is difficult or even impossible.
Like the related Nd compound, the crystal structure of Pr 3 Fe 3 Sb 7 (P6 3 /m, a = 1324.70 (5) pm, c = 419.71(2) pm at T = 296 K; Tables S1-S4, Supporting Information) consists of three structural elements. The Fe atoms are arranged in rods of face-sharing octahedra. Including the neighboring Sb atoms (Fe-Sb: 262.46(6)-297.65(5) pm), an infinite 1 ∞ [Fe 6/2 Sb 6/2 ] strand results, which resembles the 1 ∞ Mn 6/2 Si 6/2 ] strands in Mn 5 Si 3 and can be seen as condensed [Fe 6 Sb 8 ] clusters. [29] The interatomic distances in the Fe rod (257.68(6) and 259.0(1) pm) are very close to those in elemental Fe (248.2(1) pm) suggesting strong Light colored atoms at z = ¼, darker ones at z = ¾. Ellipsoids comprise 99 % of the probability density of the atoms. c) Details of the Fe and Pr partial structures. d-f) Temperature-dependent magnetization for an applied field less than 0.1 mT, 0.01 T, and 0.1 T, respectively. In (e) and (f), purple and green lines represent the orientation of the c axis parallel and perpendicular to the magnetic field, respectively, while the open and filled circles represent the data taken during cooling and heating, respectively. g) Field-dependent magnetization at 1.8 and 300 K. Filled and open circles are for the applied magnetic field parallel and perpendicular to the c axis, respectively. H a denotes the anisotropy field at 2.5 T for the applied field parallel to the c axis at 300 K. metallic bonding within the strand. [30] The second structural element is a trigonal rod 1 ∞ SbPr 6/2 ] consisting of face-sharing Pr 6 prisms, each centered by an Sb atom (Figure 1b,c). The inplane (423.49(1) pm) and out-of-plane (419.71(1) pm) Pr-Pr distances are very similar. Since they are quite large, direct interaction between Pr atoms is unlikely. However, exchange interaction could possibly be transferred through the enclosed Sb atom (Pr-Sb: 322.24(2) pm). Both structural elements extend along the c direction forming spatially separated substructures with a shortest Pr-Fe distance of 387.9(1) pm. They are interconnected by a bridging Sb atom with interatomic distances of 328.63(2) and 262.47(6) pm (Sb 2 to Pr and Fe, respectively). The 4f-3d bonding as known for RE-TM intermetallics is replaced by a 4f-5p-3d interaction in the compound. The remaining Sb atoms form planar ribbons with a diamond-shaped pattern (4 4 net), which connects the 1 ∞ Fe 6/2 Sb 6/2 ] and 1 ∞ SbPr 6/2 ] strands ( Figure S1, Supporting Information). Consistent with the structure solution, energy-dispersive X-ray spectroscopy (EDX) analysis reveals a chemical composition of 23(1) at% Pr, 23(1) at% Fe, and 52(1) at% Sb (<2 at% Bi from flux) ( Figure S2, Supporting Information).
Temperature-dependent magnetization measurements with the c axis parallel to the pick-up-coil axis show a spontaneous magnetization followed by sign reversal below 40 K at very small fields and even in the zero-field limit (Figure 1d), i.e., |µ 0 H| ≤ 10 −4 T in our setup. The sample was measured from the unpolarized, paramagnetic state, i.e., above the Curie temperature (T C ≈ 380 K) down to 1.8 K. Down to 40 K, the sample is in a state of randomly oriented, polarized domains. At about 40 K, a spontaneous magnetization emerges. Obviously, there is a spontaneous symmetry breaking that leads to a macroscopic magnetization without a compensating domain structure. We observe similar behavior for powder samples ( Figure S3a, Supporting Information).
The pseudo-one-dimensional nature of the crystal structure is also reflected in a strong magnetic anisotropy. We notice a very different temperature dependence of the magnetization for fields applied along the c axis and within the ab plane (Figure 1e,f; Figure S4a,b, Supporting Information). In the fielddependent magnetization at 300 K (Figure 1g), we observe an induced magnetic moment for small fields within the ab plane of about 4 µ B f.u. −1 , while the moment along the c axis is close to zero. This implies that the easy axis of the magnetization lies in the basal plane at this temperature. The magnetization along c reaches a plateau when the applied field is the same as the anisotropy field H a = 2.5 T. This corresponds to a reorientation of ferromagnetically ordered iron rods in the external field. At 1.8 K, on the other hand, we observe a pronounced magnetic hysteresis ( Figure S3b, Supporting Information) and a magnetic moment both for fields of µ 0 H = 0.1 T applied along the c axis (0.6 µ B f.u. −1 ) and within the ab plane (0.7 µ B f.u. −1 ; Figure 1f). These measurements evidence a canted magnetic structure at low temperatures, consistent with results presented below. The change of the easy axis provides evidence of a spin-reorientation transition (SRT). Both the c axis and ab plane isotherms at 1.8 K show coercive fields µ 0 H c of 0.07 and 0.02 T, respectively ( Figure S3b, Supporting Information). Furthermore, we attribute the pronounced maximum in the magnetization for fields applied along c at 30 K to this SRT (Figure 1e,f).
Temperature-dependent magnetization measurements in an applied field of 0.01 T along the c axis or within the ab plane show hysteresis upon cooling and heating at low temperatures ( Figure 1e). We first cooled the sample from 390 K (above T C ) to 1.8 K in zero applied field. Afterwards, we applied a magnetic field of 0.01 T upon heating up to 390 K and cooling down to 1.8 K. At 15 K, we note a fully compensated magnetization, i.e., no effective magnetic contribution. Below this compensation temperature (T comp ), we observe a magnetization reversal for the field along c. Since an applied field of 0.01 T is smaller than µ 0 H c below T comp , the magnetization fails to reorient towards the field direction resulting in an antiparallel magnetization.
We determined the H-T phase diagram in the range of the SRT ( Figure S3c, Supporting Information). For H || c, the SRT increases with field up to 3 T, above which the SRT cannot be resolved anymore ( Figure S5, Supporting Information). For H ⊥ c instead, the SRT decreases with increasing field.
Zero-field electrical-transport measurements of a Pr 3 Fe 3 Sb 7 single crystal parallel to the c axis ( Figure S6, Supporting Information) evidence metallic behavior (ρ 300 K /ρ 1.8 K ≈ 7) and a kink at ≈35 K, which corresponds to the temperature range where the SRT occurs. At low temperatures, the electrical resistivity approaches a constant value of ρ 0 ≈ 35 µ Ω cm. Figure 2 shows pulsed-field magnetization isotherms for fields up to 56 T aligned along the c axis. At 1.4 K, the magnetization increases nearly linear up to ≈40 T where it changes slope and approaches saturation at a moment of ≈12 µ B f.u.  The saturated moment of the compound is very similar at 17 K but slightly reduced at 20 K as the temperature approaches T comp . At 17 and 20 K, we find kinks in the magnetization at ≈7 and 15 T, respectively. These features are in the vicinity of the compensation temperature and the spin-reorientation transition, so changes in the spin structure are expected for such regions. At 300 K, the magnetization increases almost linearly above 3 T which we attribute to the field-induced alignment of the paramagnetic moments of the Pr atoms. The magnetic moment at highest investigated magnetic field, 56 T, is not yet representing the entire spin contributions expected for three (symmetrically equivalent) iron(I) and three praseodymium(III) ions.
To gain a microscopic understanding of the spin orientation in Pr 3 Fe 3 Sb 7 , we performed powder neutron diffraction experiments at zero applied field between 1.7 and 400 K. We found a contribution of magnetic scattering below 350 K, which is limited exclusively to the positions of the Bragg reflections of the nuclear structure (k = (0, 0, 0); Figure 3a). The refinement of the magnetic structure shows a ferromagnetic alignment of the Fe moments along the c direction with the magnetic space group P6 3 /m (no. 1374). [31] The moment increases to lower temperatures from 0.7(1) µ B at 300 K to 1.0(1) µ B at 70 K per Fe atom ( Table 1).
In the temperature range above T sr , the total magnetic moment of the compound is exclusively determined by contributions of the Fe atoms. From the neutron data, we only observe the projection of the moment on the c axis. In zerofield, there seems to be no long-range order of the moments in the ab plane. Our measurements at and below 20 K reveal an increasing contribution of the RE partial structure represented by a stepwise increase in diffraction intensity ( Figure S7, Supporting Information). Refinement in P6 3 /m results in two ferromagnetically ordered substructures that are aligned antiparallel, thus reducing the total magnetization. At 20 K, the net c contributions almost perfectly nullify, which is in accordance with the magnetization data featuring a magnetic compensation slightly below 20 K for H || c. Below T sr , the neutron data allow the refinement of canting angles against the c axis of θ Mc of 132(2)° for Pr and 37(3)° for Fe (at 7 K), but not of defined orientations (Figure 3b; Table S5 and Figure S8, Supporting Information). The canting leads to a reduction in symmetry as the mirror plane perpendicular to the c direction is lifted, changing the magnetic space group from P6 3 /m (no. 1374) to P6 3 (no. 1360). [31] We employed Mößbauer spectroscopy as a local probe technique to study the Fe spin structure in detail. 57 Fe Mößbauer spectra obtained at room temperature indicate two magnetically distinct sites exhibiting individual static magnetic hyperfine fields (Figure 4). Additionally, we need a background signal with a signal fraction of a bg = 9.1(1.1) % to describe the data. Both sample sites can be described utilizing shared linewidth ω, isomer shift IS, largest principal component of the electric field gradient (EFG) V zz , and EFG asymmetry η. The EFG was calculated from a La 3 Fe 3 Sb 7 model compound in FPLO (Table S6, Supporting Information). [32,33] The values of magnetic hyperfine field B Hyp as well as the direction of the field in the local coordinate system, θ and Φ, and, hence, the angle between the local moment and the c axis α are independent fit parameters ( Table 2). We assume the direction of the smallest principal component of the EFG (V xx ) along the c axis. This is consistent with the axial symmetry in the ab plane and η = 1. The spectral area fractions are 0.535 (12) to 0.475 (12)   Above T sr , only the spin component parallel to the c axis (grey arrow) are visible from the neutron data, probably due to disorder of the components in the ab plane (compare to Figure 5c). c) Below T sr , the Fe and Pr moments are perceptibly canted. The neutron data provides no information about the spin reorientation in the ab plane. The here shown orientations include information from Mößbauer spectroscopy (see below).  and 2, respectively. The presence of two different local hyperfine fields is clearly visible in the spectrum, resulting in the splitting of the left most peak. Figure 5a presents the temperature dependence of the magnetic hyperfine fields, following a phenomenological order-parameter behavior B = B 0 (1−(T/T C ) α ) β , while the two magnetic sites only differ in B 0 . Here β is the critical exponent describing the behavior close to T C and α is a phenomenological parameter to describe the behavior for T near zero. [34] Both sites show the same qualitative temperature dependence, only differing in B 0 . No quantitative analysis of the critical exponent is performed as no Mößbauer data is recorded near T C . This fit model can be applied at all measured temperatures above T sr ≈ 40 K. Above the spin-reorientation transition, the Mößbauer spectra are consistent with the orientation of the iron moments along the local Fe-Fe bonds (Figure 5c). Below 40 K, the two individual sites start to merge until we observe only one site at 30 K. The moments reorientate towards the center of the Fe octahedron. The hyperfine field values in this temperature range are consistent with the temperature development of site 1 above the transition, increasing with decreasing temperature and saturating at 20.6 T at 4.2 K (Figure 5a). However, changes in θ and Φ show a change in orientation of the magnetic moments relative to EFG principal axes (Table 2, right), proving the spin reorientation. The temperature-dependent isomer shift can be described utilizing a Debye fit with a Debye temperature of Θ = 380(20) K for all sites and temperatures (Figure 5b), reflecting no detectable change in the iron substructure dynamics. The 4.2 K spectrum is displayed in Figure 4b.
We measured room temperature Mößbauer spectra of a powder sample with external magnetic fields of 0, 0.1, and 1 T applied transverse to the gamma-beam direction ( Figure S9, Supporting Information). Most importantly, we observe an increase in area of the 2nd and 5th line for both applied field measurements compared to zero-field, proving a reorientation of the iron moments away from the gamma-beam direction. To rule out a possible reorientation of the powder grains, we performed a second zero-field measurement after the measurements with applied field, restoring the original zero-field spectra. Based on the magnetization and neutron data, an easy rotation of the ab component of B hyp should occur in weak external fields, while the component along the c axis should stay constant. To validate this model the area ratio of the 2nd and 5th line compared to the 1st and 6th line, k = (a 2 + a 5 )/ (a 1 + a 6 ) in applied field can be simulated based on the angle α obtained from the zero-field measurements and compared to the value obtained directly from the external field data. The area ratio expected for a given α in applied field using this model is determined by choosing randomly distributed c axis directions, as found in powder, and rotating the local moments around the c axis so that the angle between the local moment and the external field is minimal. k then can be calculated from the angles between the local moments and the gamma-beam direction. When taking the α values from the zero-field measurement α 1 = 41(5)° and α 2 = 90(10)°, we obtain k = (a 2 + a 5 )/(a 1 + a 6 ) ≈ 1.31 (α 1 ) and k ≈ 0.93 (α 2 ) for the transversal-field case. The area ratios obtained directly from the 0.1 T measurements are 1.18(9), 1.07(10) for the two sites at room temperature, consistent within error bars. When estimating α 1 and α 2 directly from the k values obtained by the 0.1 T data, we obtain α 1 = 35° and α 2 = 65°. In summary, the Mößbauer spectra measured in applied external field support the model that weak external fields of about ≈0.1 T lead to a flip of the ab spin component only. We observe no significant difference between the 0.1 and 1 T transversal-field.
Combining the findings of all our measurements, we deduce a model for the complex magnetic behavior of Pr 3 Fe 3 Sb 7 . The high temperature magnetization is governed by the Fe Adv. Mater. 2023, 35, 2207945   Figure 4. a) Mößbauer spectrum at 295 K. The data are described by two main sites (red, green) sharing V zz and IS, but using individual fit parameters for B Hyp , θ, and Φ. The blue line is the normalized sum of site 1 and 2. A background (grey) with low spectral weight (9.1 %) is indicated. b) Mößbauer spectrum at 4.2 K. The data are described by one main site (red) exhibiting a static magnetic hyperfine field. The background (grey) is fixed with the same spectral fraction. substructure, which orders ferromagnetically below 380 K in a yet unspecified domain structure. The Mößbauer data evidence that two different canted local magnetic moments exist at 40 K and above, both possessing an ab contribution. Their orientation in the crystal structure closely resembles the orientation of the Fe-Fe bonds in the Fe-strand (angles of the Fe-Fe bonds against the c axis are 35.5° and 90°). The slight increase in length for the bond along the rod-axis (257.7(1) pm) compared to the perpendicular one (259.0(1) pm) is mirrored in the observed larger site-population of site 1. The observed behavior indicates a magnetic frustration-as the spin orientation competes between in-plane and out-of-plane alignment-that can be traced back to the chemical bonding between Fe atoms. Due to the absence of long-range order of the ab spin component, neutron diffraction only displays the projection of the Fe moment on site 1 to the c axis, which mimics an easy-axis anisotropy along c. When comparing the ratio of the local hyperfine field B Hyp at room temperature and 4.2 K with the iron magnetic moments µ Fe deduced from our neutron data, we find a somewhat stronger increase in µ Fe towards lower temperatures [B Hyp (295 K)/B Hyp (4.2 K) ≈ 2/3 compared to µ Fe (300 K)/ µ Fe (7 K) ≈ 2/5]. As the model to describe the neutron diffraction data does not include spin canting above T sr , this deviation can be associated to the reduced length of the projected moment. Calculating an effective moment using the projection along c (from neutron diffraction), the population of the site and its canting angle [54 %; 41(5)°] gives 1.72 µ B per Fe (5.16 µ B f.u. −1 ). This corresponds to the spin-only value of low-spin iron(I) with the electronic configuration [Ar]3d 7 4s 0 and one unpaired electron. A similar spin-only value is also be calculated for low spin iron(III) ([Ar]3d 5 4s 0 ). However, the high oxidation state seems unlikely in the intermetallic compound. The magnetization of 5.16 µ B f.u. −1 , as obtained from Mößbauer and neutron data, is close to the magnetization at the anisotropy field observed in static and pulsed field measurements.
The local moments as observed by Mößbauer spectra are oriented with angles between 35° to 45° and 65° to 90° towards the c axis for the iron sites 1 and site 2, respectively. Applying an external magnetic field lifts the degeneracy of spin orientation in the ab plane and induces spin rotation in the direction of field, as shown in the Mößbauer data for 0.1 and 1 T. In the ab plane, spins align at much weaker fields compared to the c direction, as the in-plane rotation is a single-ion effect not hindered by neighboring atoms. However, along the c direction, columnar effects prevent an alignment, as a spin flip of an Fe atom breaks the ferromagnetic ordering with its neighbors. As a result, higher fields are required to achieve magnetic saturation.
Based on these results, we can interpret the magnetization data in more detail (Figure 1f), where the magnetization shows a decrease with decreasing temperature for H || c above T sr . This implies that, as the Fe moments increase, the internal field of the Fe substructure becomes larger and interacts with the Pr moments. At T sr , the interaction becomes sufficiently strong for the Pr substructure to order magnetically, which is reflected both in the magnetization and diffraction data. Below T sr , the refinement of the neutron data results in a canted magnetic structure as suggested by the bulk magnetization measurements and confirmed by Mößbauer spectroscopy. The neutron diffraction data point towards an antiparallel alignment of the Pr and Fe magnetic moments at low temperatures. As seen in Mößbauer spectroscopy, the onset of the Pr order causes a significant change in the iron order. A Fe-spin reorientation leads to a single canted magnetic iron site at 30 K and below, where the local moments are oriented with an angle of ≈48° towards the c axis. The coupling of the Fe and Pr moments is likely the reason for a stable refinement of the diffraction data in a noncollinear space group, above T sr the refinement converges only in the collinear model. The competing magnetic interaction is the origin for the antiparallel magnetization observed in lowfield magnetization measurements. The bulk magnetization data at 0.01 T (Figure 1e) show that the magnetic moments have a component within the ab plane even at room temperature, as visible in the data with magnetic field applied in the plane. The pulsed field experiments show a saturation magnetization of ≈12 µ B f.u. −1 at 1.4 K when all moments are aligned in field direction. This corresponds well to the magnetic moments determined by neutron diffraction, which sum up to a moment of 11.4 µ B f.u. −1 (3 (M Pr + M Fe ) = 3 (2.1+1.7) µ B f.u. −1 ).
As mentioned in the introduction, the 4f-3d interaction between light RE and TM elements typically couples the moments parallel to each other. This is a consequence of the 3d-5d hybridization in these compounds [19] that tends to lower the 3d moments. The magnetic coupling between RE and TM Adv. Mater. 2023, 35, 2207945 weakens with increasing atomic number of the RE due to the lanthanide contraction. As a consequence, the overlap between the 4f shell and 5d electrons becomes smaller and, since the 4f-5d exchange interaction is determined by this overlap, lower hybridization between the 4f shell and the 5d electrons results in reduced magnetic RE-TM interaction. [20] The spin alignment between the RE and TM elements in Pr 3 Fe 3 Sb 7 shows a different behavior. Presumably, the framework of the Sb atoms and the molecular field within the Fe substructure play a role here. The proximity of the Sb and Fe atoms may enhance the hybridization of the 3d-5d bands that transfers the spin moment from Fe to Pr through 3d-5d exchange interactions. A similar behavior was observed for La 0.75 Pr 0.25 Co 2 P 2 , where the Pr and Co magnetic moments show a ferrimagnetic alignment, an effect attributed to the P atoms modifying the 3d-4f exchange interaction. [35]

Conclusions
Our study of Pr 3 Fe 3 Sb 7 by means of bulk magnetometry, electrical transport, Mößbauer spectroscopy and powder neutron diffraction measurements demonstrate that, below 380 K only the Fe moments undergo ferromagnetic ordering. Mößbauer spectroscopy reveals two sites with spin canting of the Fe moments at room temperature, which explains the apparent contradiction between the easy-plane and easy-axis magnetization as suggested by magnetization and diffraction experiments, respectively. The origin of the Fe substructures complex behavior, still is an open question for future investigations. The magnetic moment obtained by all three techniques evidences a spin-1/2 system as in a Fe(I) lowspin configuration. At the spin-reorientation transition at about 40 K, the Pr moments order as well, thereby aligning opposite to the Fe moments. Coupling of the two substructures locks all Fe moments to a single site lifting the Fe-Fe bonding-mediated magnetic frustration. The antiparallel coupling combined with magnetocrystalline anisotropy gives rise to a magnetization antiparallel to the external field. The origin of the spontaneous symmetry breaking in very low fields is still unexplained. Nonetheless, this behavior is highly anisotropic and fragile as an increase in the external field above H c leads to sign reversal of the overall moment. This could potentially be used for magnetic switching, where temperature and external field may adjust the materials magnetic response (positive, antiparallel or fully compensated magnetic moment). Furthermore, an antiparallel orientation of the RE and TM spins is unexpected since usually the light RE spins (as for Pr) couple parallel to the TM moments. The reversal of this behavior in Pr 3 Fe 3 Sb 7 foreshadows that ferromagnetically coupled heavy RE-TM magnets could be realized. Our improved understanding of 3d-4f materials should contribute to the rational design of magnetic materials with tailored magnetic characteristics.

Experimental Section
Single-Crystal Growth: The compound was prepared starting from the elements in an argon-filled glove box (MBraun; p(O 2 )/p 0 < 1 ppm, p(H 2 O)/p 0 < 1 ppm) in 3 mL silica ampoules. The ampoules were loaded with freshly filed Pr (99.9 %, Edelmetall Recycling m&k GmbH), Fe powder (abcr, 99.90 %), and Sb flakes (reduced in H 2 stream at 450 °C). A fivefold mass excess of Bi (reduced in H 2 stream at 240 °C) was added as flux media. Silica wool was added as filter material for hot centrifugation. Evacuated and sealed ampoules were heated to 770 °C for 12 h and subsequently cooled at 1 K min −1 to 300 °C. Hot centrifugation at 500 °C was used to isolate the needle-like crystals of the target compound from the flux. Residual bismuth flux was removed by wet chemical etching in acetic acid and hydrogen peroxide (3:1). [36] Single-Crystal X-ray diffraction (SCXRD): Intensity data were collected at 296(1) K with a four-circle diffractometer Kappa Apex2 (Bruker) equipped with a CCD-detector using graphite-monochromated Mo-Kα radiation (λ = 71.073 pm). Data were corrected for Lorentz and polarization factors, [37] before applying a multiscan absorption correction. [38] The structure was solved using direct methods with ShelXT. [39] Structure refinement against F o 2 including anisotropic displacement parameters for all atoms was performed with ShelXL. [40] Further details of the crystal-structure investigation(s) may be obtained from the Fachinformationszentrum Karlsruhe, 76 344 Eggenstein-Leopoldshafen (Germany), on quoting the depository number CSD-2117929.
EDX Analysis: EDX measurements were conducted using a SU8020 (Hitachi) Scanning Electron Microscope (SEM) equipped with a Silicon Drift Detector (SDD) X-Max N (Oxford) to check the chemical composition of the crystals. Samples were prepared on a carbon pad or, for higher accuracy, embedded in epoxy resin and polished, followed by sputtering with carbon. Elemental mapping was carried out at the accelerator voltage 30 kV. EDX analysis gives a uniform elemental distribution of 23(1) at% Pr and Fe as well as 53(1) at% Sb, which is in good agreement with the sum formula obtained by SCXRD. Only negligible residues of the flux medium have been found on the surface.
Magnetization Measurements: Magnetization measurements were conducted on a powder sample and on single crystals with the magnetic field applied along either the crystallographic ab plane or the c axis using a Quantum Design superconducting quantum interference device (SQUID) magnetometer (MPMS-XL) between 2 and 400 K up to 7 T and a vibrating-sample magnetometer in fields up to 14 T. The sample length was 2.44 mm with a diameter of 0.18 mm. In view of the strong magnetic torque, the sample was glued to the holder. High-field magnetization measurements were performed using a coaxial pick-up coil system up to 56 T at the Dresden High Magnetic Field Laboratory (HLD -EMFL). [15,41] Neutron Powder Diffraction: Thermal neutron diffraction in the temperature range of 1.7-400 K was carried out on the D20 highintensity 2-axis diffractometer with a pyrolytic graphite HOPG (002) monochromator in reflection position and fixed vertical focusing at the Institut Laue-Langevin (ILL). Diffraction data were recorded to 2θ = 150° (Δ(2θ) = 0.05°) at a wavelength of λ = 241.1 pm. [42] The data were analyzed with the help of magnetic structure refinements using the Jana2006 software package [43] Electrical-Transport Measurements: Four-point resistance measurements were performed using a Quantum Design Physical Property Measurement System (PPMS) in the temperature range between 1.8 and 350 K. A single crystal of Pr 3 Fe 3 Sb 7 was placed on top of a SiO 2 lamella having triangular sheets of copper as current leads. As voltage contacts, 99.9 % high-purity Au wires of 25 µm diameter were used fixed transversally on the sample with graphite paste.
Mößbauer Spectroscopy: 57 Fe Mößbauer measurements were conducted in a nitrogen-shielded Cryo Vac liquid Helium flow cryostat. A Rh/Co source driven by a Mößbauer WissEL drive unit MR-360 biased by a DFG-500 frequency generator in sinusoidal mode was used. The detection device was a proportional counter tube in combination with a CMTE multichannel data processor MCD 301/8K and a WissEL timing single channel analyzer (SCA) to set the energy window. Two samples of Pr 3 Fe 3 Sb 7 powder were investigated. All data evaluation has been performed using the program Mössfit employing transmission integral fits to accord for sample thickness. [44] The electric field gradient (EFG) principal components are defined with |V zz | ≥ |V yy | ≥ |V xx | using an asymmetry parameter η = (V xx − V yy )/V zz . θ is the angle between the magnetic hyperfine field (B Hyp ) direction and V zz . Φ denotes the angle of the projection of B Hyp onto the V xx /V yy plane and V xx . α is the angle between the local crystallographic c axis and B Hyp , ω signifies the half width at half maximum (HWHM) absorber linewidth. All isomer shifts (IS) are stated relative to α-Fe. The measurements have been carried out at temperatures from 4.2 to 300 K.
Quantum Chemical Calculations: Calculations of the EFG were performed with the FPLO package [32,33] based on a scalar relativistic, closed shell DFT calculation with the PW92 functional [45] using a k-mesh of 8 × 8 × 22 points.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.