Among the common first-row transition metal ions, Ni 2+ stands out as one of the relatively rare examples of a d 8 S = 1 system. This is important because the presence of strongly correlated S = 1 magnetic values unlocks a wide range of nonclassical magnetic behavior. In particular, it makes Ni 2+ systems potential candidates for so-called Haldane behavior. 1 Haldane behavior contributed to the awarding of the Nobel Prize in Physics in 2016 and postulates that in antiferromagnetic one-dimensional (1-D) substructures, ions with a formal odd integer spin state can induce the formation of a ground singlet state with an energy gap between that of the classical S = 1 triplet state. This singlet state is the product of nonclassical quantum behavior, and the energy of the gap between this singlet ground state and the classical triplet is the so-called Haldane gap. [2][3][4] In this case, the presence of magnetic frustration in the chains prevents the formation of magnetic ordering to very low temperatures, while the nonmagnetic singlet ground state is a candidate for a highly degenerate quantum spin liquid ground state. 5 This is very different from the classical behavior of S = 1/2 ions in 1-D structures. Despite the importance of this observation and the interest it generated, there have been relatively few Haldane materials observed experimentally. Shortly after Haldane published his initial work in this area, Sakai and Takahashi built upon Haldane's ideas finding three other possible magnetic phases in the vicinity of the Haldane phase for S = 1 systems. 6 These various phases are a function of the ratio between interchain magnetic coupling constants and the single ion anisotropy of the metal ions within the chains.
Previously we have been successful in the synthesis of a wide range of compounds featuring parallel quasi-1-D chains based on first-row transition metal ions and magnetically silent tetrahedral oxyanion building blocks ((VO 4 ) 3-, (MoO 4 ) 2-). 7,8 The tetrahedral oxyanions play an important role in structure design as they can coordinate to the metal ions in a wide variety of ways and act as both intra-and interchain bridging groups. 9 They also play an important role in intra-and interchain magnetic coupling. By using a high-temperature hydrothermal method, namely hydrothermal reactions between 300 and 650 °C, we can often obtain high-quality single crystals of a wide range of materials with parallel quasi-1-D chains. 10 The single crystals play an important role in that they are often large enough to be selectively aligned in the magnetic field for magnetic measurements. This is particularly important for this low-dimensional class of compounds since they are highly anisotropic, and this anisotropy is very important in understanding the physical properties of the materials. Thus, the ability to orient single crystals in magnetic fields is critical to understanding their anisotropic magnetic structures.
In this vein, we recently undertook the investigation of reactions of Ni 2+ ions with various oxyanions to explore the structural phase space and obtain large single crystals for magnetic studies. In this paper, we report results utilizing phosphates as the tetrahedral oxyanion building block. Phosphates are quite common metal-containing minerals but are still somewhat underdeveloped as a building block in synthetic metal ion chains. A recent study of nonstoichiometric ellenbergerite-type phosphates suggests the synthetic chemistry and magnetic properties in these systems is quite rich. 11 Fortunately, metal phosphates are also well suited for the hightemperature hydrothermal approach. Phosphates also offer an entry point to then explore the direct isoelectronic comparison to arsenates and vanadates, which surprisingly often display quite different magnetic behavior. 12 Herein we report three new structural types featuring quasi-1-D nickel oxide chains with various phosphate bridging behavior, Ni 2 (PO 4 )(OH) (I), Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 (II), and NaNiPO 4 (III). These compounds exhibit complex structures and varied magnetic behavior. In particular, we observe that one of the compounds, Ni 2 (PO 4 )(OH), displays Haldane-like behavior and this complex magnetism is discussed in some detail.
2.1. General Materials and Methods. Compounds were synthesized via a high-temperature hydrothermal technique in sealed silver ampules within Tuttle-seal autoclaves. 13 Commercially available reactants were used as received without further purification: CaHPO 4 (Strem, 99.95%), K 2 HPO 4 (EMD Chemicals, 98%), NaH
2 PO 4 •H 2 O (Thermo scientific, >99%), NiF 2 (Advanced Research Chemicals Inc., 99%), NiO (Strem, 99%), CsOH (Strem, 99.9%), RbOH (Strem, 99.9%), KOH (Alfa Aesar, 99.98%), and NaOH (Acros Organics, 98%). 2.2. Synthesis of RbH 2 PO 4 . RbH 2 PO 4 was synthesized in house using commercially available Rb 2 CO 3 (Alfa Aesar, 99%) and H 3 PO 4 (Aldrich, 85 wt %) via the reaction below. 14 + + + Rb CO 2H PO 2RbH PO CO H O 2 3 3 4 2 4 2 2
Rb 2 CO 3 (0.87 g) was dissolved in water and allowed to stir for 30 min. Phosphoric acid (0.44 mL) was then added dropwise to the solution which was stirred for another 30 min. Lastly, methanol was added dropwise until precipitation appeared complete, and the
reaction was stirred for another 30 min thereafter. 14 The solid product was rinsed with acetone and dried in an oven at 80 °C for at least 2 h. The product was confirmed to be RbH 2 PO 4 (Figure S1) by powder X-ray diffraction (PXRD) and stored in a dry desiccator. Synthesis of CsH 2 PO 4 (Figure S2) was accomplished using a similar reaction scheme using Cs 2 CO 3 (Alfa Aesar, 99%).
Compound I was prepared from the hydrothermal reaction of CsH 2 PO 4 (0.1408 g, 0.61 mmol) and NiF 2 (0.0592 g, 0.61 mmol) in a 1:1 molar ratio with 0.4 mL of 1 M CsOH mineralizer for 7 days at 650 °C. Millimeter long bright green columnar crystals formed in quantitative yield after filtration and water wash (Figure 1a).
Compound II was prepared from the hydrothermal reaction of CaHPO 4 (0.1476 g, 1.1 mmol) and NiF 2 (0.0419 g, 0.43 mmol) in a 2.5:1 molar ratio with 0.4 mL of 1 M KOH mineralizer for 14 days at 650 °C. Half millimeter long pale green columnar crystals formed (40% yield) (Figure 1b) along with a white powder identified via PXRD to be fluorapatite, Ca 10 (PO 4 ) 6 F 2 and fluorite, CaF 2 , 45% and 15% respectively, (Figure S6). The desired product was easily separated manually.
Compound III was prepared from the hydrothermal reaction of NaH 2 PO 4 •H 2 O (0.1233 g, 0.89 mmol) and NiO (0.0771 g, 1.0 mmol) in an approximate 1:1 molar ratio with 0.4 mL of 1 M NaOH mineralizer over 7 days at 600 °C. Millimeter-sized yellow elongated block-like crystals formed in quantitative yield (Figure 1c).
2.6. X-ray Diffraction. Single crystal X-ray diffraction intensity data were collected at room temperature on a Bruker D8 Venture diffractometer. Data were collected using phi and omega scans of 0.5°w idth. Data were integrated (SAINT) and corrected for absorption (SADABS) within the APEX3 software suite. 15 The structures were solved using SHELXT (intrinsic phasing) and refined using full matrix least-squares techniques on F 2 with SHELXL. 16 All non-hydrogen atoms were refined anisotropically. Hydrogen atoms in the structures of Ni 2 (PO 4 )(OH) and Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 were located from the difference electron density map in combination with bond valence sum considerations of their host oxygen atoms, and their positions refined. The hydrogen atom on the (HPO 4 ) 2-group of Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 was set to 1/6 occupancy due to symmetry considerations. The structures of Ni 2 (PO 4 )(OH) and Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 were refined in the polar space groups P31m and P6 3 mc, respectively. The correct absolute structures were supported by their respective Flack parameters of 0.02(2) and 0.007 (12). Details of the structure refinements are given in Table 1. Selected bond lengths are provided in Table 2. Structural data were deposited with the joint CCDC/FIZ Karlsruhe deposition service, CSD deposition 2334932-2334934. Bulk phase identification was performed by PXRD using a Rigaku SmartLab diffractometer with Cu Kα radiation (λ = 1.5406 Å). Data were measured in 0.01°increments at a scan speed of 5°/min from between 5 and 80°in 2θ. Comparisons to the calculated patterns from the single crystal structure refinements were made using the JP-Minerals VESTA software package. 17 2.7. Magnetic Property Characterization. DC magnetometry was performed on a Quantum Design DynaCool Physical Property Measurement System (PPMS) equipped with a 9-T magnet. A coalignment of single crystals was utilized for Ni 2 (PO 4 )(OH) and NaNiPO 4 to obtain a sufficient mass of sample for high-quality data. An array of seven aligned crystals was affixed to a clean quartz paddle by rubber cement such that the applied external field, H, would be parallel to the long growth axis in both cases. Respective aligned crystal masses were 0.64 mg and 0.50 mg. Given the smaller size of the Ni 7 (PO 4 ) 3 (HPO 4 ) 3 single crystals, a 10.5 mg powder sample was used. The powder was loaded into a supplied holder from the PPMS VSM Option kit and fitted into a brass trough. Temperaturedependent magnetism (M-T) was measured in zero-field cooled (ZFC) and field-cooled (FC) modes from 2 to 300 K with a 2 K/min sweep rate in various applied fields up to 90 kOe. Isothermal fielddependent magnetism (M-H) was performed at fixed temperatures below and slightly above points of interest. Potential hysteresis was assessed by sweeping ±90 kOe with ΔH = 90 Oe/s (0.1% |H max |). Diamagnetic contributions were subtracted as calculated by Pascal's constants and background subtraction of rubber cement (∼10 -10 emu/Oe/mg cement). 18 Heat capacity measurement of a 0.28 mg single crystal of Ni 2 (PO 4 )(OH) oriented with H perpendicular to the long column axis was performed using the semiadiabatic method of the Heat Capacity PPMS User Option. This orientation was chosen to provide the maximal sample contact area with the heat sink platform. Apiezon N grease was used to affix the crystal to the sample platform mount.
2.8. Additional Characterization. Infrared (IR) spectroscopy was used to confirm the presence of the hydroxyl groups in the
Table 2. Selected Interatomic Distances (Å) in Ni 2 (PO 4 )(OH), Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 , and NaNiPO 4 Ni 2 (PO 4 )(OH) Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 NaNiPO 4
Ni 2 (PO 4 )(OH) and Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 materials. Crystalline powder samples were ground for a uniform mixture and used to collect the IR spectrum using the Shimadzu IRAffinity-1S Infrared Spectrometer. The infrared spectrum was collected over 400-4000 cm -1 with a resolution of 8 cm -1 . Elemental analysis was performed by energy dispersive X-ray analysis (EDX) with a Hitachi SU6600 scanning electron microscope using a Schottky (FE) source.
Semiquantitative data from this technique confirmed the absence of a fluoride ion in both the Ni 2 (PO 4 )(OH) and Ni 7 (PO 4 ) 3 (HPO 4 )-(OH) 3 crystals, further supporting the hydroxide assignment in the crystal structures, and in harmony with the infrared spectra.
3.1. Synthesis and Crystal Growth. Crystals of Ni 2 (PO 4 )(OH) (I) and Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 (II) formed over a range of reaction conditions and mineralizer solutions. The phases were sometimes intermixed depending on the reaction conditions, so powder diffraction was used to optimize the synthetic conditions and to verify the purity of samples used for magnetic measurements. In the reactions using Rb + or Cs + counterions, we found that despite the use of NiF 2 as the nickel source, their final composition showed no evidence of fluoride incorporation in the lattice from EDX analysis. We do, however, observe considerable sensitivity to reaction temperature and the identity of the alkali metal counterion on the final product distribution, regardless of whether that ion is incorporated in the products or not. Thus, we performed a systematic investigation of reactions with various alkali metal ions. Reactions using a 1:1 molar ratio of NiF 2 , RbH 2 PO 4 , and 1 M RbOH produce a mixture of I and II together at 650 °C, while the same reaction at 600 °C forms a pure powder of II. Using Rb + reagents generally did not lead to very large crystals and variation of reaction conditions did not substantially improve size. The anisotropic nature of the structures, however, encouraged us to pursue the growth of larger single crystals to enable alignment in fields. We found that using Cs + ions at 650 °C led to sizable single crystals of I (1 mm in length). It should be noted that when K 2 HPO 4 and 1 M KOH were employed, a novel fluoride-containing material, K 2 Ni 2 (P 2 O 7 )F 2 , was produced, which will be discussed in a future publication. When we employed an alkaline earth phosphate source, CaHPO 4 , we were gratified to isolate sizable single crystals of II up to 0.5 mm in length, albeit with various calcium-containing powder side products. The improved size of the crystals of II, however, was sufficient to mechanically separate a phase pure sample. Upon changing the phosphate source to NaH 2 PO 4 •H 2 O and again using NiF 2 as the nickel source, the product was the previously reported phase Na 2 Ni(PO 4 )F (Figure S11). 19 Changing the nickel source to NiO to remove fluoride from the synthetic equation naturally caused the product distribution to change significantly, producing NaNiPO 4 crystals (III) in a phase pure yield, with sizes exceeding 1 mm in length. The phosphate groups provide additional connectivity both within and between the three-chain clusters. The clusters are templated by the P(1) phosphates that sit in the center of the three chains. The P(2) phosphate connects three oxygen atoms of one cluster to one oxygen atom of a neighboring cluster. The P(3) phosphate connects to three clusters at its O(9) vertex, and to individual oxygen atoms of those clusters
through the basal oxygen atoms of the tetrahedron. The nickel oxide chain and the unidirectional orientation of the phosphate tetrahedra impart the polar axis of the structure. The overall topology of the structure is related to that of holtedahlite-type (and by extension, the ellenbergerite-type) structures of Mg 12 (HPO 4 ,PO 4 )(PO 4 ) 5 (OH,O) 6 . 20 Considering this, the current formula of Ni 2 (PO 4 )(OH) could be expanded to Ni 12 (PO 4 ) 6 (OH) 6 , which reflects the full hydroxide occupancy on the nickel oxide chains of this structure type, as supported by the bond valence sums (Table S1). In This chemical formula is known across a range of divalent metals and the various transition metal compounds with the formula M 2 (PO 4 )(OH) (M = divalent transition metal) exhibiting several structure types, including the trigonal structure type observed for I in the present study, though to our knowledge no Ni 2+ analog has been reported for any of these polymorphic forms. Among the polymorphs reported for Co 2 (PO 4 )(OH), the trigonal polymorph has been given the γdesignation, with αand βforms crystallizing in Pnnm adamite-type and I4 1 /amd caminite-type structures, respectively. 21 Spin-glass behavior was observed at low temperatures in γ-Co 2 (PO 4 )(OH). 21 The other M 2 (PO 4 )(OH) and M 2 (AsO 4 )(OH) analogs reported in the structural literature appear to be confined to the orthorhombic and tetragonal polymorphs. [22][23][24][25][26][27][28][29][30][31][32][33][34] The difficulty in isolating purely nickel-based analogs of these compounds has been documented elsewhere, 35 so their synthesis as reasonably large-sized single crystals from high-temperature hydrothermal reactions is particularly fortuitous.
Compound II is a novel ellenbergerite-type structure crystallizing in the polar space group P6 3 mc. The structures of several ellenbergerite, phosphoellenbergerite, and arsenoellenbergerite analogs have been reported from natural and synthetic samples. 11,21,22,36,37 The structure of II features a nickel oxide chain substructure ( pseudohexagonal hole in II that is occupied by a second chain of face-sharing nickel oxide octahedra involving Ni(2) with Ni•••Ni = 2.4927(3) Å (Figures 5 and 6). This independent chain runs parallel to the more complex Ni(1) chains and is linked by phosphates to the Ni(1) chains. This relationship w a s a l s o r e c e n t l y o b s e r v e d i n t h e s t u d y o f Na 2x Co 6 (OH) 3 (HPO 4 )(H x/3 PO 4 ) 3 and its relationship to γ-Co 2 (PO 4 )(OH). 23 However, in the structure of II, the hexagonal channels host the independent Ni(2) chains instead of Na + ions. Another notable difference between II and I is the presence of the (HPO 4 ) 2-group that templates the three-chain clusters in the former. This (HPO 4 ) 2-tetrahedron shows a more significant difference in basal P-O and apical P-OH bond lengths (1.505(6) Å versus 1.561(13) Å, respectively) that was not present in the analogous phosphate tetrahedron in I (basal P-O = 1.534(6) Å, apical P-O = 1.508(13) Å). Additionally, the hydrogen atoms originating from the bridging oxygen atoms of the nickel oxide chains are oriented more favorably toward the basal oxygen atoms of the (HPO 4 ) 2-group in II, in contrast to I where they form O-H•••O interactions to the apical oxygen atom of the phosphate groups. As such, the hydrogen atom of the (HPO 4 ) 2-group of II was assigned to a one-sixth occupied 12d Wyckoff site attached to the apical oxygen atom that sits on a 2b Wyckoff site. Here, the hydrogen atom of the (HPO 4 ) 2-group forms weak O-H•••O interactions with the three-chain clusters (O-H = 0.832(6) Å; H•••O = 2.531(8) Å; O•••O = 3.37(2) Å, O-H•••O = 178.86(13)°). Again, a 3-fold expansion of the chemical formula of Ni 2 (PO 4 )(OH) to Ni 6 (PO 4 ) 3 (OH) 3 can thus accommodate the addition of a charge-balanced Ni(HPO 4 ) unit (Ni 2+ from the face-sharing chains at the center of the hexagonal channels) to form Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 .
The unit cell volume expands by about 25% to accommodate this additional component as expected, primarily via expansion of the a-and b-axis lattice parameters as the hexagonal channels are introduced. The (HPO 4 ) 2-units and the nickel oxide chain substructures again exhibit a polar orientation, now with hexagonal symmetry instead of trigonal symmetry. The occurrence of II and I under similar synthetic conditions in our hands may also reflect their close structural relationship. In general, almost all members of the ellenbergerite structural family are notorious for displaying significant nonstoichiometry in their crystal lattice, so the ability to prepare stoichiometrically precise materials opens the class for more detailed physical property measurements. 11,21 3.2.3. Structure of NaNiPO 4 (III). Crystals of III appear to be a new monoclinic polymorph of this NaNiPO 4 formula, compared to the more extensively characterized orthorhombic maricite-type structure, often studied for battery applications. 38,39 To our knowledge, this monoclinic structure type (space group P2 1 /c) of hydrothermally grown NaNiPO 4 single crystals in the present study has also not been reported for any other analogous of NaMPO 4 (M = divalent transition metal) compositions for which several other structure types have been reported, including olivine, maricite, and stuffed tridymite. [40][41][42][43][44][45][46][47][48][49] In particular, the monoclinic structure types of NaCoPO 4 (P2 1 /c, 5-coordinate Co 2+ ; P2 1 /n, ABW-type), β- NaCuPO 4 (P2 1 /n, highly distorted CuO 6 units), and NaZnPO 4 (P2 1 /n, beryllonite-type; P2 1 /n, ABW-type) have significant structural differences from monoclinic III. [42][43][44][45][46][47] The structure of the hydrated NaNiPO 4 •7H 2 O phase is likewise significantly different, reported in space group P4 2 /mmc. 50 The structure of monoclinic III in the present study has two unique nickel sites forming one-dimensional nickel oxide chains built from edge-and corner-shared NiO 6 distorted octahedra (Figure 7 6) to 170.28 (7)°about Ni (2). Geometries about the phosphate groups are typical for those isolated tetrahedra.
The magnetic behavior of I proved to be very intriguing. Temperature-dependent susceptibility (Figure 8) shows a broad hump-like curvature from room temperature down to 20 K giving way to various low-temperature orderings below 10 K. No dependence on cooling mode (ZFC/FC) was seen. This broad rollover for a quasi-1-D S = 1 chain is characteristic of a Haldane-like material. [2][3][4] An anomaly at 32
K (more visible in the inset plot) is also present, hinting at what is believed to be an intrachain ordering between Ni 2+ centers (see heat capacity discussion below). The low-field paramagnetic tail below the anomaly is suppressed upon increasing the applied field (inset Figure 8), a feature complemented by the initial ferromagnetic saturation-like curvature of 2 and 5 K isotherms in field studies. The transition to a linear regime in the isothermal magnetization curves >20 kOe is attributed to a low-field reorientation of the paramagnetic Ni 2+ spins to an ordered one. The ordered state appears to exist primarily intrachain based on the residual susceptibility with coupling existing between the two unique Ni 2+ dimer-like building blocks of each columnar chain. Further, subtle features were seen in the 50 and 90 kOe susceptibilities indicative of magnetic substructures. Heat capacity measurements of a single crystal in applied fields of both 0 and 10 kOe, although somewhat noisy due to its small size, clearly demonstrate a cusp-like feature at 31 K above the estimated linear phonon baseline (Figure 9). This temperature matches the subtle inflection point of the susceptibility plot suggesting an antiferromagnetic ordering event.
A number of attempts were made to fit the broad Haldanelike curve using several models generally based on the classical Bonner-Fisher approximation of quasi-1-D chains. 51,52 We found that the Padéapproximation approach, like that employed for the fitting of the SrNi 2 (VO 4 ) 2 Haldane material by Bera et al., 53 could successfully model the data from 35 to 300 K (Figure 9). Attempts to fit the model in the lowtemperature regime saw degradation of the fit due to the suspected antiferromagnetic ordering anomaly near 32 K but did capture the general shape of the curve with the downturn and subsequent Curie tail from the paramagnetic Ni 2+ chain ends (Figure S12). The equation and calculated constants for fitting an S = 1 quasi-1-D Heisenberg antiferromagnetic chain are given by Law et al. 54 An excellent fit to the broad feature was obtained with the constants collected in Figure 9. The inclusion of a Curie-like addition to the model was needed to account for the singular unpaired spins at the ends of Ni 2+ chains. Using the free ion assumption with S = 1/2 and g imp = 2.002, this Curie-like "impurity" was found to be marginal at only 3.53(6)% contribution of the fitted range. The chain-end impurity Curie-Weiss temperature, Θ imp , of -44(1) K captures the suspected antiferromagnetic behavior of the 32 K anomaly. The fitted nearest-neighbor intrachain exchange coupling constant, J NN , is 7.993(3) meV (J NN /k B = 92.76(3) K). The fitted g-factor of 2.30(2) computes to an effective moment of 3.25 μ B , which is higher than the nominal spin-only moment but typical for most Ni 2+ ions in distorted nonideal O h symmetry. 55,56 Haldane materials require an odd integer spin value in quasi-1-D chains, inclusion of an S = 1 Ni 2+ oxidation state, and longrange antiferromagnetic ordering. 2,4,57 The magnetic ordering occurs when each odd integer spin on a site fractionalizes, leaving the sites having an odd number of S = 1/2 spins that behave independently. 5 In turn, the spins form the magnetic equivalent of resonance valence bonds which in turn create singlets in the magnetic moment generating a gapped, nonmagnetic ground state. 57,58 This process is a gradual function of temperature, generating the characteristic broad rollover curve. This nonmagnetic ground state is known as the Haldane gap (Δ) and is proportional to the magnetic coupling constant, J/k B (K). 3 It is believed that I could be a Haldane gap material after considering required qualifications including, ideally, an inelastic neutron scattering experiment that will provide both the anisotropic value D and the interchain magnetic exchange coupling constant J, whose ratios will provide the nature of the Haldane-like material.
A related material previously reported to display Haldane behavior, SrNi 2 (VO 4 ) 2 , consists of one-dimensional nickel spiral chains linked by magnetically silent vanadate tetrahedra. 53,59 The nearest-neighbor intrachain interactions, as determined through a series of computational and neutron scattering studies, were found to take place along a diagonal via a complex Ni-O-V-O-Ni pathway through the empty dorbitals of the vanadium ion (Figure 10). 59 When examining the vanadate to phosphate connections we can see differences between the two structures. For SrNi 2 (VO 4 ) 2 , the vanadate tetrahedra connect the otherwise isolated nickel chains together. In I, however, this is not the case. The structure contains edge-shared zigzag chains of distorted octahedra, but these couple to form dimeric ribbon-like chains. These ribbonlike chains also corner-share via oxygen atoms to create a nearest-neighbor interactions. This is not the case in SrNi 2 (VO 4 ) 2 where only one unique nickel site is present and the nickel oxide polyhedral edge-share via ∼180°to one another creating a dimer set that is then edge-shared to the next set by ∼90°creating a helical chain. We think that the multiple magnetic interactions in I are responsible for the complex magnetic behavior we observe.
3.3.2. Magnetism of Ni 7 (PO 4 ) 3 (HPO 4 )(OH) 3 (II). Unfortunately, single crystals of II did not grow large enough in sufficient quantity to enable alignment in a magnetic field, so the magnetic studies were performed on well-ground powders of the single crystals. The powder magnetization studies of II demonstrated a primarily paramagnetic system that undergoes ferromagnetic ordering at approximately 55 K as field strength is increased (Figure S13). Compound II possesses the most uniform Ni 2+ octahedral bond lengths of the three materials we examined in this work and the magnetization was perhaps the most straightforward. This may be due to the powder acting as an ensemble average of crystallographic orientations and diluting any moderate anisotropic behavior. Further, the S = 1 Ni 2+ has weaker spin-coupled interactions in MO 6 octahedra versus other 3d metals Co 2+ (S = 3/2) and Mn 2+ (S = 5/2). 60 The curvature of M-H isotherms below 20 K gave additional evidence for this para-to ferromagnetic transition with increasing field strength. No saturation was observed at 2 K when subject to fields up to 90 kOe, suggesting that stronger fields are required to encourage any antiferromagnetic interchain behavior. Such studies would also be aided by orientable single crystals given the possibility of highly anisotropic interactions down the c-axis chains. The trigonal arrangement along the Ni 2+ dimer chains offer a distinct possibility for anisotropic magnetic frustration should antiferromagnetic coupling between chains be present.
Curie-Weiss analyses in 10 and 50 kOe fields demonstrated respective para-and ferromagnetic behavior (Figure 11). When fitting the data over the 200-300 K range, both showed similar behavior. In the 10 kOe ZFC setting, Curie-Weiss temperatures, Θ, of -168(1) K suggests magnetic interactions are predominately antiferromagnetic with evidence for slight antiferromagnetic behavior in the 50-150 K range yielding to ferromagnetic interactions below 50 K. The effective magnetic moment, μ eff , of 3.45(4) μ B , as computed from the respective Curie constant for the 10 kOe curve, was higher than the expected for the ideal spin-only moment of 2.83 μ B for d 8 Ni 2+ but is not surprising for Ni 2+ complexes. 55,56,61 3.3.3. Magnetism of NaNiPO 4 (III). Magnetic investigations of III show antiferromagnetic behavior for all tested fields (10, 50, 90 kOe) with no noteworthy difference between ZFC and FC cooling modes (Figure S14). A cusp-like feature indicative of the antiferromagnetic phase transition T N was identified near 18(1) K. There is an inflection point at 10 K that could be related to the net ferromagnetic interchain ordering of the Ni 2+ chains, though subsequent isothermal M-H yielded only linear curves up to 90 kOe. Fitting susceptibility data to the Curie-Weiss law in the 200-300 K paramagnetic region (Figure 12) yielded a Θ of -13.5(1) K. The Curie constant of 1.33(1) cm 3 -K/mol equated to an effective magnetic moment of 3.26(5) μ B . The difference between this experimentally observed moment and the ideal spin-only effective moment of 2.83 μ B is suggested to be the result of low energy 3 T 2g states from nominally octahedral symmetry. In this case, low-lying 3 T 2g states can mix with the 3 A 2g ground state, creating some spin-Orbit coupling with concomitant increase in the effective moment above the ideal spin-only value. 61 A similar behavior was noted in II as described above and is well-known in classical Ni 2+ complexes. 55,56,61
Three new Ni 2+ phosphates have been synthesized using a high-temperature hydrothermal method and characterized structurally and magnetically. All three structures consist of 1-D chains of distorted Ni 2+ octahedra that are edge-or faceshared in various ways to form parallel chains. The chains are all linked by corner sharing tetrahedral phosphates. In all cases, the structures are well-ordered and stoichiometric on all sites. In compounds containing hydroxide or protonated oxyanion building blocks such as these, a hydrothermal synthetic method is attractive because classical high-temperature ceramic methods can lead to dehydration and formation of alternate structures. In many previous cases related to compounds described herein, they are reported as nonstoichiometric, highly defected materials. This is particularly true for the ellenbergerite structural class, which is notorious for nonstoichiometric behavior. 21 The high-temperature hydrothermal method employed in this paper generally leads to fully stoichiometric products with little or no observable defects, disorder, or partial site occupancy. This is very helpful in the accurate determination of physical properties such as magnetism. The method also enables the growth of large high-quality single crystals. This is an important aspect of this project because the highly anisotropic nature of the structures can be a vital component of the nonclassical quantum behavior of the systems. Parallel 1-D chains often display complex magnetic coupling with competitive behavior within and between the chains that can lead to magnetic frustration and nonclassical quantum behavior.
One of the compounds, Ni 2 (PO 4 )(OH) (I) displays a strong indication of Haldane-like behavior with the characteristic broad rollover of magnetic moment over a hundred Kelvin, followed by sharp antiferromagnetic ordering. The true nature of this apparent Haldane-like behavior cannot be absolutely confirmed until exchange coupling and anisotropic D values can be determined by neutron scattering. The magnetic behavior is also quite complex near and below the antiferromagnetic transition, with an unusual inflection point near 32 K and field-dependent post-transition behavior hinting at the onset of additional magnetic phenomena. The fit for the potential Haldane-like candidate, Ni 2 (PO 4 )(OH), is remarkably similar to that found with the known Haldane SrNi 2 (VO 4 ) 2 when both are oriented with the Ni 2+ chain Additional plots of M-T at various static fields and isothermal M-H can be found in the Supporting Information (Figure S14).
axes parallel to the applied field. 53 Additional work employing inelastic neutron scattering, additional magnetic investigations, and computational studies of exchange interactions to better understand this complex magnetic structure, as well as determination of the potential Haldane gap is underway. The ability to grow large (≥1 mm) single crystals will allow us to orient the crystals in external fields and obtain data related to anisotropic magnetic behavior.
The other two materials reported here, despite containing parallel quasi-1-D chains of S = 1 Ni 2+ , do not show any evidence of Haldane-like behavior. Rather they display more conventional ferro-and antiferromagnetic ordering for II and III, respectively, although both provide some indication of competitive magnetic coupling both between and within chains. The magnetic behavior of all three compounds shows considerable sensitivity to external applied magnetic fields. In the cases of II and III the chain structures are especially complex. In the case of II, there are two completely different 1-D Ni 2+ chains running parallel to each other, while in III there is only one type of chain but contains two different unique Ni 2+ ions with very different structural environments. We postulate that these multiple environments might contrive to quench the quantum frustration that leads to Haldane behavior. Nevertheless, this work demonstrates that new nickel phosphates can be synthesized as well as ordered, high-quality single crystals capable of displaying a wide range of complex magnetic behavior including potential quantum effects like a Haldane gap. Further detailed magnetic and neutron scattering study of nickel phosphates as well as exploration of Ni 2+ ions with other tetrahedral oxyanion building blocks is underway.
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.4c00872.
Powder diffraction patterns; infrared spectra; EDX elemental analysis; table of bond valence sums for I; table of Na-O bond lengths for III; additional magnetic data (PDF)
https://doi.org/10.1021/acs.inorgchem.4c00872Inorg. Chem. 2024, 63, 13265-13277