Sodalite-like Rare-Earth Carbonate : A Study of Structural Transformation and Diluted Magnetism

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains. Accepted Manuscript Dalton Transactions


Introduction
Open framework materials with porous structures, especially with zeolitic topologies, have attracted considerable attention for their permanent porosity, excellent thermal and chemical stabilities, and are widely used in industrial applications. 1 Among the 229 zeolitic topologies, the traditional zeolite is built exclusively from TO 4 tetrahedra, with the T atoms being from the main group or transition metal group, such as Si 4+ , Al 3+ , Ge 4+ , Zn 2+ , etc. 2 Rare earth (RE) elements are ideal candidates to incorporate into open framework materials for the exploration of new applications, since they contain unique properties in photoluminescence, magnetism and so on. 3 Several microporous lanthanide silicates that are built from SiO 4 tetra-hedra and LnO n (n ≥ 6) polyhedra have been successfully prepared. 4 Moreover, RE ions have become attractive candidates for constructing new single-ion magnets (SIMs) as well as singlemolecular magnets (SMMs), because most of them have a large unquenched orbital angular momentum, which may bring significant anisotropy to the system. 5 Synthetic efforts along this line have led to the discovery of many materials exhibiting SIM behaviour, expanding to lanthanide clusters and even complexes with 3-dimensional (3D) metal-organic framework structures. 6 However, few magnetic studies are focused on RE inorganic materials with 3D open framework structures. 7 The synthesis of traditional open framework materials is typically carried out in a gel medium under solvothermal conditions by using alkali metal ions or organic amines as templates or structure-directing agents (SDAs). 8 However, this strategy is not suitable for the synthesis of RE open frameworks, as the RE ions have variable and high coordination numbers as well as poor directionality, which is significantly different from transition metal ions. 9 For the design of RE open frameworks, it is worthwhile to refer to the synthesis of transition or lanthanide-based clusters, 10 in which the carbonate (CO 3 2− ) anion is an extremely versatile bridging ligand. Aside from balancing the positive charges of metal ions (3d or 4f ), each oxygen atom of CO 3 2− anion in the clusters may act as a monodentate, bifunctional, or trifunctional ligand to generate carbonato-bridged com- † Electronic supplementary information (ESI) available: Crystal data and structure refinement and selected bond lengths [Å] for 1-3 and 1-c; thermal ellipsoids for 1-3 and 1-c; 3-dimensional structure of 1-c; PXRD patterns for all samples; TGA curve for 1; XPS for 1 and 1-c; temperature dependence of M vs. H, in-phase (χ′) and out-of-phase (χ″) ac susceptibility curves of 1, 2, 1d, 2d, 1-c and 1d-c; crystallographic data for 1-3 and 1-c (CIF). CCDC 1416221-1416224 for 1-3 and 1-c. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5dt03314d plexes. 11 The CO 3 2− anion affords various coordination modes, ranging from simple chelation to more intricate coordination modes (μ 3 , μ 4 , μ 5 and μ 6 ) to construct complicated metal-carbonate cores. 12 Therefore, polycarbonate-lanthanoid cores with diverse structures are highly appropriate to construct special RE compounds with 3D open framework topologies. 13 The CO 3 2− anion may also act as a SDA in the design of RE compounds with unique structures, since the CO 3 2− anion in the carbonates can be easily removed by calcinations.
Herein we present a series of novel rare earth carbonates RE 3 (OH) 6 (CO 3 )Cl (RE = Dy, 1; Er, 2; Y, 3) that exhibit the sodalite-like structure constructed from the building unit (BU) of [RE 4 (µ 3 -OH) 4 ] cubes by introducing an appropriate amount of CO 3 2− from Na 2 CO 3 or atmospheric carbon dioxide. Remarkably,

Syntheses
Compounds 1, 2 and 3 were prepared by the hydrothermal reaction of a mixture of RECl 3 ·6H 2 O, 2.5 M NaOH, HCl (37 wt%) and H 2 O to get a pH value of 6. Finally, Na 2 CO 3 was introduced to make a reaction gel of 1.0RECl 3 : 1.2HCl : 2.5NaOH : 0.33Na 2 CO 3 : 332H 2 O. The gel was added into a 12 mL Teflonlined stainless steel autoclave and reacted at 200°C for 72 h. Many parameters including the pH value and reaction temperature were found to affect the formation of RE 3 (OH) 6 (CO 3 )Cl. The mixture of HCl and NaOH was used to reach pH = 6 and provide excessive chloride that is beneficial for a high yield.
Notably, the addition of Na 2 CO 3 is essential for the formation of the product. It has been previously shown that the structures of RE clusters and cages are able to be modified by incorporating CO 3 2− anions as a template, which is critical for the further aggregation and growth of the product. 14 In this system, the reaction without Na 2 CO 3 leads to another RE compound with a 2D structure, RE(OH) 2 Cl, as well as a few target crystals. It suggests that the CO 3 2− anion is necessary to generate the 3D structure in this reaction, and it may be introduced from atmospheric carbon dioxide. Contrarily, if Na 2 CO 3 is added in a quantity greater than the stoichiometric ratio of 0.33, another known RE compound RE(OH)CO 3 is obtained, and the yield of this by-product increases. By further increasing the ratio of Na 2 CO 3 to 1, the pure phase of the by-product can be produced. Hence the presence of an appropriate amount of CO 3 2− is vital for the syntheses of 1, 2 and 3.

Crystal structures
Single-crystal structure analyses reveal that 1, 2 and 3 crystallize in the cubic space group Im3, Im3m and Im3m, respectively (Tables S1-S3, ESI †). The asymmetric unit for each compound contains 0.5 crystallographically distinct RE sites and 0.16667 crystallographically distinct C sites (Fig. S1 anions located in the centre of the 6-ring opening and Cl − anions in the channels to achieve charge balance (Fig. 1a). As shown in Fig. 1b, the Dy atoms are linked with each other via six μ 3 -OH − , and each Dy atom belongs to two [Dy 4 (OH) 4 ] units. The [Dy 4 (OH) 4 ] unit is the only BU of the framework. In the centre of the 6-ring opening, a μ 6 -CO 3 2− ligand binds the six dysprosium ions via a μ 6 -η 2 :η 2 :η 2 bridging mode, by coordinating every O atom of CO 3 2− to two Dy atoms on the 6-ring. This coordination mode of CO 3 2− is common in clusters. 12d Because of the high symmetry of the structure, the O atoms here are disordered in two places, each one having an occupancy of one half. The six BUs on the 6-ring opening are linked with each other by vertex-sharing and they form a geo- metry similar to that of {Ni 12 } or {Mn 12 } clusters, which are also templated by CO 3 2− anions. 15 Furthermore, an SOD cage can be observed through the vertex-sharing of 24 BUs (Fig. 1c). The cage shows six 4-ring openings and eight 6-ring openings, which are composed by four BUs and six BUs, respectively. Each 6-ring opening is sealed by a CO 3 2− anion. Cl − anions in the cages act as the template here, which is similar to the role of alkali metals or organic amines in traditional zeolites. Interestingly, the structure of the super cage is the same as a 60-metalatom cluster, [Er 60 (L-thre) 34 (µ 6 -CO 3 ) 8 Br 12 (ClO 4 ) 18 (H 2 O) 40 (L-thre = L-threonine), that also possesses a sodalite cage structure featuring 24 vertex-sharing [Er 4 (µ 3 -OH) 4 ] cubes. 16 The cores of the 60-metal-atom clusters, as well as the {Ni 12 } or {Mn 12 } clusters, are encapsulated in a coordination sphere, making them independent. But in this framework, the cages combine with each other to give rise to a 3D structure. In general, the anions, such as CO 3 2− and Cl − , may have a significant function in directing the structure of rare earth complexes. Besides, the triangular coordination geometry of CO 3 2− may induce competing magnetic interactions between the metal centres, such as geometrical spin frustration, which is believed to be beneficial for enhancing magnetocaloric effects. 17 The Dy (Fig. 2f ). The framework of 1 shows the topology of SOD, as shown in Fig. 2g. The free diameters are 4.1 × 4.3 Å 2 (Dy⋯Dy distance) for the 6-ring channel, which is comparable to those of 8-ring channels in aluminosilicate zeolites, such as CHA, LEV, LTA, etc. The framework density (FD) of 1 is 12.3 Dy atoms per 1000 Å 3 . That is lower than SOD with FD Si of 16.7 T/1000 Å 3 and comparable to some zeolites with 8-ring channels due to the long distance of the Dy-O bond. 18 The SOD topology was firstly discovered for the mineral sodalite in 1930, and used to be studied in zeolites. 19 Recently, the SOD topology has also been exhibited in several ZMOFs or ZIFs. 20 This is one of the rare examples of a rare earth complex with a zeolite topology. 21 It may act as a guidance to explore new RE compounds with zeolite topology by using a [RE 4 ] tetrahedron as building unit.
The framework of 1 can be described as a three-period net with the highest symmetry of Im3m containing only one type of T site as determined by the Systre software. 22 The vertex symbol for the T site in the net is 3 6 ·4·5 4 ·6 4 . This three-period  , the framework of 1 also can be described as a three-period net of SOD topology with the highest symmetry of Im3m. The vertex symbol for the T site in the net is 4 2 ·6 4 . This three-period net is carried by a unique natural tiling with a transitivity of (1 1 2 1). There is one type of tile in this tiling with a face symbol of [4 6 ·6 8 ]. TG analysis has been performed to study the thermal stability of 1. It shows a total weight loss of 18.6% from 200 to 800°C with two stages with the first one occurring between 300-450°C (Fig. S3 †). Under heat treatment at 370°C for 4 h, the structure will transform into 1-c. Surprisingly, the structure of 1-c can be solved using single-crystal X-ray diffraction. This phenomenon is rarely observed in RE inorganic compounds. In the analysis of the single-crystal X-ray diffraction data, we noticed that the CO 3 2− anions are removed. The space group has transformed into trigonal R3 for 1-c from cubic Im3 for 1. Besides, some micro variation occurs in the structure. Here, we consider the CO 3 2− anions are removed by forming CO 2 during the calcinations. From the TG curve, the first stage weight loss of 13.6 wt% occurs at around 370°C, which corresponds with the removal of CO 2 and H 2 O. The complex then has a molecular formula of Dy 3 O 4 Cl. The possible chemical equation is shown as follows:

Dalton Transactions Paper
The PXRD pattern of 1-c corresponds well with the simulated one based on the single-crystal structure of 1-c, indicating the pure phases of the as-synthesized sample (Fig. S4 †). The formula of 1-c is confirmed by ICP, elemental analyses and anion chromatography analysis (Dy, 82.7; C, 0.41; H, 0.92; Cl, 5.95; Anal. Calcd Dy, 83.0; C, 0; H, 0; Cl, 6.04 wt%). In addition, X-ray photoelectron spectroscopy (XPS) of 1 and 1-c are also measured to verify the existence of Cl. Fig. S5 † shows the XPS survey spectra and narrow scan XPS spectra of 1 and 1-c. The signal at around 198 eV is consistent with the binding energy of Cl 2p, indicating the existence of Cl − in the structures. 23 The content ratios of Dy/Cl for 1 and 1-c derived from the XPS analysis are close to 3, corresponding to the formula of Dy 3 (OH) 6 (CO 3 )Cl and Dy 3 O 4 Cl, respectively. Additionally, the calcination derivatives (2-c and 3-c) with similar structures can be obtained following the same treatment for 2, and 3.
Single-crystal structure analysis of 1-c reveals that the asymmetric unit contains three crystallographically distinct Dy sites and four crystallographically distinct O sites (Fig. S6, ESI †). Dy(1) and Dy (2) (2)) to form anion-centered ODy 4 tetrahedra, which is common in inorganic compounds. 24 (Fig. 3c). As shown in Fig. 3d, the high-resolution TEM images (HRTEM) of 1-c clearly show a hexagonal lattice with void distances of 5.9 Å, which corresponds to the distances of the crystallographic planes of (021), (201) and (221) (5.883 Å). This further confirms the structure of 1-c.

Magnetic analyses
Dc magnetic susceptibility data of 1, 1-c and the corresponding diluted samples (1d and 1d-c, respectively) are measured in the temperature range of 2-300 K (Fig. 4). At 300 K, the χ m T products for the samples are 13.90, 13.81, 13.29 and 13.39 cm 3 K mol −1 for 1, 1d, 1-c and 1d-c, respectively, which are compatible with the calculated value of 14.17 cm 3 K mol −1 for the ground state of one Dy III ion (4f 9 , S = 5/2, L = 5, J = 15/2 and g = 4/3) in the free ion approximation. Upon cooling, the χ m T products decrease gradually at first and then more rapidly to the minimum at 2 K for all of the compounds. The χ m T versus T plots for compounds 2 and 2d are recorded between 2 and 300 K under a dc field of 500 Oe and 1 kOe, respectively (Fig. 4). The χ m T values at room temperature are 10.87 and 10.80 for 2 and 2d, respectively, which are close to the expected value of 11.48 cm 3 K mol −1 for one Er III ion (4f 12 , S = 3/2, L = 6, J = 15/2 and g = 6/5). In the high-temperature range, the χ m T product decreases slowly upon cooling for both compounds; below 75 K, the χ m T product drops quickly to a minimum value of 3.74 cm 3 K mol −1 at 2 K for 2, and a minimum value of 6.17 cm 3 K mol −1 for the diluted sample of 2d.
The magnetization versus field plots are determined at low temperatures in the range of 0-70 kOe, as shown in Fig. S8. † For 1 and 1-c, the magnetization at 2 K increases continuously upon an increase of the magnetic field, reaching the maxima of 5.57N β and 5.04N β for 1 and 1-c, respectively. These obtained values are far from the saturated magnetization for one Dy III ion, but are in agreement with the data reported pre-viously as a result of the large magneto-anisotropy and/or the low lying excited states of the Dy III ion. For 2, similar behavior of the field-dependent magnetization is observed, with a highest value of 4.46N β at 2 K, which again does not exhibit saturation, possibly due to the presence of large magnetic anisotropy.
Alternating current (ac) susceptibility measurements were performed on the microcrystalline samples to investigate the dynamic behaviour of the magnetization at low temperatures, as displayed in Fig. S9. † For 1, 1-c and 2, curves based on the in-phase and out-of-phase signals of the ac susceptibility are all superimposed on a nearly single-master curve, either in the absence of an external dc field or at an applied dc field of 1 kOe. This phenomenon often exists in high-nuclearity lanthanide cages, and is probably due to the crowded distribution of metal ions in high-nuclearity clusters, as well as in titled complexes. 25 Magnetic dilution may enhance the slow magnetic relaxation U eff originating from SIM behaviour in mononuclear and polymeric compounds, such as 1D dysprosium chains, 2D layers, and even 3D LiYF 4 compounds. 26 It is an advantage here that we can study the magnetism of samples with very large dilution ratios of up to 0.227%, for which the magnetic signal is still precluded by the instrument sensitivity for 1 and 1-c, while the same cannot be detected for most diluted samples reported. Therefore, the doped samples 1d and 1d-c in Dy : Y molar ratios of 1 : 440 are prepared to examine the nature of the magnetism. As shown in Fig. 5, ac measurements for both diluted samples in zero dc field show obvious frequency dependence below 8 K, an indication of slow relaxation of the magnetization. Unfortunately, no maximum is observed in both the in-phase and out-of-phase signals of the ac susceptibility, possibly due to the presence of a fast relaxation of the magnetisation through a quantum tunnelling mechanism. Thus, ac measurements are taken under an applied small external dc field of 500 Oe to mitigate the quantum tunnelling (Fig. S10 †). Significant frequency-dependent χ′ and χ″ signals Fig. 4 Experimental χ m T products of 1, 1d, 1-c, 1d-c, 2 and 2d in an applied field of 2 kOe, 1 kOe, 1 kOe, 1 kOe, 0.5 KOe and 1 KOe for 1, 1d, 1-c, 1d-c, 2 and 2d, respectively. The χ m T curve has been rescaled for one RE III ion. appear for both compounds and a maximum in χ′ is clearly observed moving to higher temperatures towards higher frequencies especially for 1d, indicating SIM behaviour. For 2, a similar phenomenon has occurred (Fig. S11 †). The doped sample 2d with a Dy : Y molar ratio of 1 : 15 shows slow relaxation of its magnetization under a dc field of 1 kOe, proving the position effect that the dilution of open framework materials has on the SIM behaviour once again. All the information points out that such a slow relaxation is likely dominated by single ion effects, and vanishes when more of the lanthanide ions are incorporated. The magnetic exchange interaction, although it is weak between lanthanides, and the dipolar interaction, as well as the orientation of the easy axes of magnetization work together in the molecular architecture, leading to a complicated situation which is not beneficial to the SIM behaviour.

Experimental section
Reagents and solvents were commercially available and used as received without further purification.

Synthetic procedures
Synthesis of Dy 3 (OH) 6 (CO 3 )Cl (1). Typically, dysprosium(III) chloride hexahydrate (DyCl 3 ·6H 2 O, 1 mmol) was dispersed into 5 mL distilled water, and fully dissolved by adding 0.1 mL hydrochloric acid (37 wt%, 1.2 mmol). Then 1 mL NaOH solution (2.5 mol L −1 ) was added to the mixture with stirring for 1 h to get a pH value of 6. Finally, Na 2 CO 3 was introduced to make a reaction gel of 1.0DyCl 3 : 1.2HCl : 2.5NaOH : 0.33 Na 2 CO 3 : 332H 2 O. About 1 h later, the gel was added into a 12 mL Teflon-lined stainless steel autoclave and heated at 200°C for 72 h, followed by cooling to room temperature for 20 h. The white pure crystals were obtained by washing in distilled water several times and dried at room temperature overnight after filtration. Yield: 176 mg (77% based on Dy). Elemental analysis of the product gave (in wt%) Dy: 70.9 (Calcd Dy: 71.2), C: 1.84 (Calcd C: 1.75), H: 1. 21 (Calcd H: 0.876), and anion chromatography analysis gave (in wt%) Cl: 5.11 (Calcd Cl: 5.18), which were consistent with the elemental content calculated from the molecular formula of Dy 3 (OH) 6 (CO 3 )Cl.
Synthesis of Dy 0.0068 Y 2.9932 (OH) 6 (CO 3 )Cl (1d, 0.227% diluted sample of 1). DyCl 3 ·6H 2 O/YCl 3 ·6H 2 O (in a 1 : 449 molar ratio; 10 mmol in total) was dispersed into 50 mL water, and 1 mL hydrochloric acid (37 wt%, 12 mmol) was added. Then 10 mL NaOH (2.5 mol L −1 ) was added to the mixture with stirring for 1 h to get a pH value of 6. Finally, Na 2 CO 3 was introduced to make a reaction gel of 0.002DyCl 3 : 0.998YCl 3 : 1.2HCl : 2.5 NaOH : 0.33Na 2 CO 3 : 332H 2 O. About 1 h later, the gel was added into a 100 mL Teflon-lined stainless steel autoclave and heated at 200°C for 72 h, followed by cooling to room temperature for 20 h. Pure crystals of 1d were obtained after washing the product in distilled water and were dried at room temperature overnight. The Dy/Y ratio of the product based on ICP analysis (in wt%) Dy: 0.238, Y: 57.2, gave the molecular formula Dy 0.0068 Y 2.9932 (OH) 6 (CO 3 )Cl.
Preparation of Dy 3 O 4 Cl (1-c, the calcined sample from 1). The as-synthesized 1 was placed into a normal furnace and heated from RT to 370°C at a temperature ramp rate of 1°C min −1 in air with a 4 h isothermal hold at 370°C before cooling to room temperature at a rate of 1°C min −1 . The transformation of the structure occurred during this process. The composition of 1-c is studied by ICP, elemental analyses and anion chromatography analyses (Dy, 82.7; C, 0.41; H, 0.92; Cl, 5.95; Anal. Calcd Dy, 83.0; C, 0; H, 0; Cl, 6.04 wt%). Singlecrystal analysis show that 1-c exhibits a new structure. Besides, the calcination derivatives (2-c and 3-c) with a similar structure can also be obtained following the same treatment of 2, and 3. The structure and purity of the samples are confirmed by PXRD.
Preparation of Dy 0.0068 Y 2.9932 O 4 Cl (1d-c, 0.227% diluted sample of 1-c). The as-synthesized 1d was placed into a normal furnace and treated following the method similar to 1-c. The structures of 1d-c are confirmed by PXRD.

Characterization methods
PXRD data were collected on a Rigaku SmartLab(3) diffractometer with Cu Kα radiation (λ = 1.5418 Å). Inductively coupled plasma (ICP) analysis was performed on a Perkin-Elmer Optima 3300DV spectrometer. Elemental analysis was conducted on a Vario MICRO elemental analyzer. Anion chromatography analysis was performed on a DX-500 Ion Chromatography (IC) System of Dionex. Thermogravimetric analysis (TGA) was carried out on a TA Q500 analyzer in air with a heating rate of 10°C min −1 from RT to 800°C. X-ray photoelectron spectroscopy (XPS) measurements were performed using a Thermo Escalab 250 spectrometer with monochromatized Al Kα excitation. The TEM and HRTEM images were taken on a JEOL JEM-3010 transmission electron microscope operating at an accelerating voltage of 300 kV. Magnetic measurements were performed on a Quantum Design MPMS XL-7 SQUID magnetometer on polycrystalline samples. Diamagnetic corrections were made with Pascal's constants for all the constituent atoms and sample holder. Alternating-current (ac) magnetic susceptibility data were collected on the same instrument employing a 3.5 Oe oscillating field at frequencies up to 1442 Hz.

X-ray crystallographic analyses
Suitable single crystals of Dy 3 (OH) 6 (CO 3 )Cl (1), Er 3 (OH) 6 (CO 3 )Cl (2), Y 3 (OH) 6 (CO 3 )Cl (3) and Dy 3 O 5 Cl 0.5 H 1.5 (1-c) with dimensions of 0.32 × 0.21 × 0.15, 0.07 × 0.07 × 0.05, 0.05 × 0.05 × 0.05 and 0.06 × 0.06 × 0.03 mm 3 , respectively, were selected for single-crystal X-ray diffraction analyses. The intensity data were collected on a Bruker Apex II DUO area-detector diffractometer with graphite-monochromated Mo Kα radiation (λ = 0.71073 Å) at a temperature of 23 ± 2°C. Cell refinement and data reduction were accomplished with the SAINT processing program. 28 The structures were solved by direct methods and refined by a full matrix least-squares technique with the SHELXTL crystallographic software package. 29 The heaviest atoms of Dy, Er, Y, O, Cl and C could be unambiguously located. The H atoms in the structures of 1, 2 and 3 were added theoretically, and that of 1-c were not added. All nonhydrogen atoms were refined anisotropically. CCDC 1416221-1416224 contain the supplementary crystallographic data for this paper. Crystal data and refinement parameters for the structure determination are presented in Tables S1-3 and S5. † Selected bond distances are listed in the Table S4. †