Controlled Zn(II) to Co(II) Transmetalation in a Metal–Organic Framework Inducing Single-Ion Magnet Behavior ^†

Abstract

The intrinsic characteristic features of metal–organic frameworks offer unique, great opportunities to develop novel materials with applications in very diverse fields. Aiming to take advantage of these, the application of post-synthetic methodologies has revealed itself to be a powerful approach to the isolation and structuration of metal ions, molecules, or more complex species, either within MOF channels or reticulated at their network, rendering novel and exciting MOFs with new or improved functionalities. Herein, we report the partial post-synthetic metal exchange of Zn(II) metal ions by Co(II) ones in water-stable three-dimensional CaZn₆-MOF 1, derived from the amino acid S-methyl-L-cysteine, allowing us to obtain two novel MOFs with increasing contents of the Co(II) ions Co_4%@1 and Co_8%@1. Remarkably, the presented post-synthetic metal exchange methodology has two relevant implications for us: (i) it allowed us to obtain two novel MOFs, which were not accessible by direct synthesis, and (ii) enabled us to transform physical properties within this family of isoreticular MOFs from the diamagnetic pristine MOF 1 to MOFs Co_4%@1 and Co_8%@1, exhibiting field-induced, frequency-dependent, alternating current magnetic susceptibility signals, which are characteristic features of single-molecule magnets.

1. Introduction

Single-ion magnets (SIMs) represent the ultimate limit of the miniaturization of single-molecule magnets (SMMs) [1,2,3,4,5,6,7,8,9,10]. Their potential applications as high-density magnetic memories and quantum computing devices in the emerging field of molecular spintronics [11,12,13,14,15] have attracted the attention of many groups working in the fields of molecular magnetism [16,17,18,19,20] and multifunctional magnetic materials [14,21,22,23,24]. In fact, during the last decade, great advances have been achieved in efforts to increase phase memory times and the temperature need to observe magnetic bistability [25,26,27,28]. SIMs have been reported with very distinct metal ions, including lanthanides [8,29] and actinides [30,31], but also with first- [32,33,34] and third-row [35,36] transition metal ions. An interesting family of SIMs are Co(II) complexes. Their inherently large magnetic anisotropy, due to first-order spin–orbit coupling (SOC), allow for the obtention of Co(II)-based SIMs with different coordination geometries and signs of magnetic anisotropy [32,37,38]. However, despite the commendable advances achieved in the field, several daunting challenges remain to be solved, with the controlled spatial organization of magnetic molecules to build solid ordered arrays while retaining their intrinsic individual magnetic properties being one of the greatest ones.

Metal–organic frameworks (MOFs) [39] have emerged as a novel generation of porous materials [40,41] due to their unique differentiating features, such as their modulability and rich host–guest chemistry [42] and high porosity and crystallinity [43,44]. Since their advent, MOFs have shown noteworthy applications in very diverse fields, from gas adsorption and separation [45,46] and catalysis [47,48] to drug delivery [49] and magnetism [50,51], recently being used, for example, in water remediation [52] and chemical nanoreactors as well [53]. This, in part, is a direct consequence of the commendable synthetic efforts to achieve great control of the MOF structure obtained. In fact, even though total control is not always possible to achieve, the different pre- and post-synthetic methodologies [54,55,56,57] used have made it possible to precisely tune the ratio of distinct components and to gain control of MOF dimensionality and topology and, consequently, of the MOF’s final properties.

Thus, MOFs, a priori, represent outstanding frameworks for taking serious steps forward toward gaining control of the spatial organization of magnetic molecules. Indeed, seminal works have been devoted to this end, where the main synthetic strategies can mainly be classified into the following: (i) the direct self-assembly of metal ions with selected polytopic ligands, including mixtures of two distinct metal ions (doping) [37,58,59], (ii) the encapsulation of preformed SMMs [60,61], and (iii) post-synthetic metal exchange [62]. Regarding this last approach, we have already successfully performed it. We used a diamagnetic Ca^IICo^III 3D MOF as the initial framework, and, post-synthetically, we replaced all Ca^II ions with different lanthanide metal ions, where the final MOFs with Kramers ions (Dy^III and Er^III) exhibited SIM behavior [24]. Also, recently, partial post-synthetic metal exchange was reported in a Zn-based MOF, replacing Zn(II) with Co(II) in different, very low atomic percentages [62]. Here, underpinned in our treasured knowledge of MOFs and SIMs, we further explore the latter route by performing partial post-synthetic metal exchange in a diamagnetic heterobimetallic CaZn₆-MOF (Figure 1), replacing Zn(II) with Co(II) metal ions in different controlled atomic percentages (4 and 8%), and we study how this different degree of metal heterogeneity influences the magnetic properties of the obtained Ca^IIZn^IICo^II-MOF. The resulting novel mixed-metal MOFs exhibit field-induced, frequency-dependent, alternating relaxation typical of single-molecule magnets.

2. Materials and Methods

2.1. Materials

All reagents were obtained from commercial sources (Merck-Aldrich, Darmstadt, Germany) and used without further purification unless otherwise indicated. Reactions were performed in conventional round-bottomed flasks or sealed vials equipped with a magnetic stirrer. {CaZn₆[(S,S)-Mecysmox]₃(OH)₂(H₂O)}·12 H₂O (1) was prepared following a previously reported procedure [63].

2.2. Preparation of {CaCo_0.24Zn_5.76[(S,S)-Mecysmox]₃(OH)₂(H₂O)}·14 H₂O (Co_4%@1) and {CaCo_0.48Zn_5.52[(S,S)-Mecysmox]₃(OH)₂(H₂O)}·16 H₂O (Co_8%@1)

Polycrystalline powders of 1 (0.2 g, 0.12 mmol) were suspended in water solutions in two independent vials with different amounts of Co(NO₃)₂·6 H₂O (0.01 and 0.10 g for Co_4%@1 and Co_8%@1, respectively) for 1 h under mild stirring. Then, the resulting solids were isolated by filtration, thoroughly washed with water and methanol, and air dried. Co_4%@1: Anal.: calcd for C₃₀H₆₈N₆O₃₅S₆CaCo_0.24Zn_5.76 (1689.5): C, 21.32; H, 4.05; S, 11.08; N, 4.84%. Found: C, 21.37; H, 3.99; S, 11.12; N, 5.01%. Co_8%@1: Anal.: calcd for C₃₀H₇₂N₆O₃₇S₆CaCo_0.48Zn_5.52 (1735.9): C, 20.75; H, 4.18; S, 11.52; N, 5.03%. Found: C, 20.83; H, 4.09; S, 11.56; N, 5.11%.

2.3. Physical Techniques

Elemental (C, H, S, and N) and ICP-MS analyses were performed at the Microanalytical Service of the Universitat de València. The thermogravimetric analysis was performed on crystalline samples under a dry N₂ atmosphere with a Mettler Toledo TGA/STDA 851e thermobalance, manufactured by Mettler Toledo, Greifensee, Switzerland, operating at a heating rate of 10 °C min^–1. A scanning electron microscopy (SEM) elemental analysis was carried out for 1, Co_4%@1, and Co_8%@1, using a HITACHI S–4800 electron microscope, sourced from HITACHI High-Technologies Corporation, Tokyo, Japan, coupled with an energy dispersive X-ray (EDX) detector. Data were analyzed with the QUANTAX 400 system manufactured by Bruker AXS, Karlsruhe, Germany.

2.4. Gas Adsorption

The N₂ adsorption–desorption isotherms at 77 K were carried out on polycrystalline samples of 1, Co_4%@1, and Co_8%@1 with a BELSORP-mini-X instrument manufactured by BEL Japan, Inc., Osaka, Japan. The samples were first activated with methanol and then evacuated at 343 K during 16 h under 10^–6 Torr prior to their analysis.

2.5. X-Ray Powder Diffraction Measurements

Polycrystalline samples of 1, Co_4%@1, and Co_8%@1 were introduced into a 0.5 mm borosilicate capillary prior to being mounted and aligned on an Empyrean PANalytical powder diffractometer, manufactured by PANalytical, Almelo, The Netherlands, using Cu Kα radiation (λ = 1.54056 Å). Five repeated measurements were collected at room temperature (2θ = 2–45°) and merged in a single diffractogram.

2.6. Magnetic Measurements

Direct current (dc) magnetic susceptibility measurements (2–300 K) under applied dc magnetic fields of 5000 G (T ≥ 20 K) and 2500 G (2 ≤ T ≤ 20 K) and variable-field (0–5 T) magnetization measurements (2.0–10 K) on finely crushed crystals of Co_4%@1 and Co_8%@1 (mixed with grease to avoid the crystallite orientation) were carried out with a Quantum Design MPMS-XL7 SQUID magnetometer, manufactured by Quantum Design, San Diego, CA, USA. Variable-temperature (2–8 K) alternating current (ac) magnetic susceptibility measurements for Co_4%@1 and Co_8%@1 were performed under an applied dc field of 1 and 2.5 T, a ±5.0 G oscillating field, and at frequencies between 0.1 and 10.0 kHz with a Quantum Design Physical Property Measurement System (PPMS) manufactured by Quantum Design, San Diego, CA, USA. The susceptibility data were corrected for the diamagnetism of both the sample holder and the constituent atoms.

The magnetic properties of octahedral cobalt(II) complexes are usually dominated by a first-order spin–orbit coupling (SOC). The coupling between the spin (S = 3/2) and orbital momentum (L = 1)—using the T-P isomorphism [64]—in a distorted octahedral environment generates six Kramers doublets. Among these, the two lowest-energy doublets are usually the only states populated below 300 K. Consequently, a phenomenological approach based on a zero-field splitting (ZFS) of a state S = 3/2, considering only these doublets, was employed. In the spin Hamiltonian (Equation (1)), the parameters D and E represent the axial and rhombic magnetic anisotropy, respectively.

$${\widehat{H}}_{GS}=\beta H\left(g_x{S}_x+g_y{S}_y+g_z{S}_z\right)+D\left({\widehat{S}}_z^2-{\displaystyle \frac{S\left(S+1\right)}{3}}\right)+E\left({\widehat{S}}_x^2-{\widehat{S}}_y^2\right)$$

(1)

The activation energy (Ea) in Co_4%@1 and Co_8%@1, based on the phenomenological approach employed, was evaluated using the equation ${ E}_a=2\sqrt{D^2+3E}$, where the value of E value is always positive.

3. Results and Discussion

3.1. Synthesis and Characterization

In this work, we present the solid-state post-synthetic partial metal exchange of Zn(II) metal ions by Co(II) ones, in a previously reported water-stable three-dimensional MOF, derived from the amino acid S-methyl-L-cysteine, with the formula {CaZn₆[(S,S)-Mecysmox]₃(OH)₂(H₂O)}·12 H₂O (1) (Mecysmox = bis[S-methylcysteine]oxalyl diamide), allowing us to obtain two novel MOFs with distinct contents of Co(II) ions of formulas {CaCo_0.24Zn_5.76[(S,S)-Mecysmox]₃(OH)₂(H₂O)}·14 H₂O (Co_4%@1) and {CaCo_0.48Zn_5.52[(S,S)-Mecysmox]₃(OH)₂(H₂O)}·16 H₂O (Co_8%@1). Interestingly, direct synthesis attempts—adapting the previously reported procedure for 1 [63]—using the oxamidate-based ligand H₂Me₂-(S,S)-Mecysmox in a basic methanolic solution with mixtures of aqueous solutions of Co(NO₃)₂·6 H₂O, Zn(NO₃)₂·6 H₂O and CaCl₂ were unsuccessful in synthesizing Co_4%@1 and Co_8%@1.

Co_4%@1 and Co_8%@1 were obtained by immersion of the pristine polycrystalline powder of 1 in aqueous solutions containing different amounts of Co(NO₃)₂·6H₂O (0.01 and 0.10 g for Co_4%@1 and Co_8%@1, respectively) for 1 h under mild stirring. This process was carefully monitored visually, confirming the absence of powder dissolution and reprecipitation. The nature, pureness. and identity of Co_4%@1 and Co_8%@1 were conclusively established through a combination of physical techniques, including elemental analyses (C, H, S, and N), inductively coupled plasma mass spectrometry (ICP-MS), scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDX), powder X-ray diffraction (PXRD), nitrogen adsorption isotherms at 77 K, and thermogravimetric analyses (TGAs) (see Supporting Information). ICP-MS and SEM-EDX enabled us to determine the atomic percentages of metal ions in each MOF (Tables S1 and S2), and together with elemental analyses and TGAs allowed us to establish their chemical formulas (see Section 2). A comparison of ICP-MS and SEM-EDX data for 1, Co_4%@1, and Co_8%@1 confirmed that only Zn(II) ions in 1 have been replaced with Co(II) ions, since the atomic ratios between M(II) and Ca(II)—where M = Zn for 1 and Zn + Co for Co_4%@1 and Co_8%@1—remained unchanged. These results also allowed the quantification of the percentage of substituting Co(II) ions in each compound. We determined Co(II) contents of 4 and 8%, relative to Zn(II)ions, in Co_4%@1 and Co_8%@1, respectively. SEM micrographs and EDX elemental mapping confirmed a homogeneous distribution of Zn(II), Ca(II), and Co(II) ions in 1, Co_4%@1, and Co_8%@1 (Figures S1–S3). The TGAs of Co_4%@1 and Co_8%@1 (Figure S4) revealed a rapid mass loss from room temperature to ca. 100 K, followed by a plateau until decomposition started at ca. 300 K. The calculated percentage weight loss values at 150 K were 14.5% for Co_4%@1 and 16.8% for Co_8%@1, corresponding to the release of 14 and 16 water molecules, respectively. The PXRD analyses evidenced a match between the theoretical pattern of 1, derived from previously reported single-crystal X-ray diffraction data, and the experimental patterns of each sample (Figure 2). This confirmed the purity of Co_4%@1 and Co_8%@1, as well as their isoreticular nature with respect to 1. Additionally, the preservation of permanent porosity in Co_4%@1 and Co_8%@1 after the post-synthetic partial metal exchange was demonstrated by means of nitrogen adsorption isotherms at 77 K (Figure S5). All samples exhibited fully reversible type I isotherms, characteristic of microporous materials with permanent microporosity (Figure S5). The estimated Brunauer–Emmett–Teller (BET) surface areas [706 (1), 694 (Co_4%@1), and 686 m² g⁻¹ (Co_8%@1)] as well as the calculated pore widths [0.6622 (1), 0.6387 (Co_4%@1), and 0.6589 nm (Co_8%@1)] (Figure S6) were nearly identical, confirming the preservation of permanent porosity. Interestingly, the presented post-synthetic partial metal exchange leads to a change in the magnetic properties of the pristine MOF 1, resulting in new MOFs of the oxamidate family, which exhibit SIM behavior (see magnetic properties below).

3.2. Description of the Structure

Although the crystal structure of the parent MOF 1 has been previously reported, it is worth describing it here, particularly to understand the possible coordination environments and geometries of the Co(II) ions substituting those of Zn(II) in compounds Co_4%@1 and Co_8%@1, which, in turn, are responsible for the SIM behavior. MOF 1 crystallizes in the chiral P6₃ space group and consists of a chiral honeycomb three-dimensional (3D) calcium(II)–zinc(II) network (Figure 1) featuring hexagonal channels. The network is constructed from trans-oxamidato-bridged dizinc(II) units {Zn^II₂[(S,S)-Mecysmox]} (Figure 1a), which act as linkers between Ca(II) cations through the carboxylate groups and are further interconnected to neighboring dizinc(II) moieties by water–hydroxide groups (Figure 1a). The Zn(II) ions are five-coordinate, exhibiting a trigonal bipyramidal geometry. They coordinate one amidate nitrogen atom, one carbonyl oxygen atom, one carboxylate oxygen atom, one carboxylate oxygen atom from a neighboring dizinc(II) unit, and one water–hydroxide bridging group (in a 1:2 statistical distribution). Thus, a priori, the potential coordination environment of substituting Co(II) ions should be the same as that of the replaced Zn(II) ones. However, with a lack of crystal structure, it is difficult to establish their precise coordination environment, being possible other environments, such as octahedral. Each calcium(II) ion is nine-coordinate in a distorted monocapped square antiprismatic geometry (CaO₉), built by six carboxylate oxygen atoms from six oxamidato groups and three oxygen atoms from water–hydroxide groups (Figure 1).

3.3. Magnetic Properties

3.3.1. Static (dc) Magnetic Properties

The dc magnetic properties of compounds Co_4%@1 and Co_8%@1, as the χ_MT against the T plot, are shown in Figure 3, with χ_M being the normalized dc molar susceptibility per one cobalt(II) per formula unit (the raw χ_MT against the T plot is shown in Figure S7). The χ_MT(T) data for Co_4%@1 and Co_8%@1 are typical for a high-spin d⁷ Co(II) site. The χ_MT value at room temperature is 2.52 cm³ K mol^–1 for Co_4%@1 and Co_8%@1, which is consistent with previously reported mononuclear cobalt(II) complexes with unquenched orbital momentum contributions [37,65]. Upon cooling, χ_MT decreases smoothly down to 1.42 (Co_4%@1) and 1.56 (Co_8%@1) cm³ K mol^–1 at 2 K, revealing the existence of a significant ZFS arising from the second-order spin–orbit coupling (SOC) of the Co(II) ion (Figure 3). This was further supported by the M vs. H plots, with magnetization values of 1.71 (Co_4%@1) and 1.95 Nβ (Co_8%@1) at 3 K and 5 T (Figure 4). These values are well below the saturation magnetization for a high-spin d⁷ Co(II)ion (M_sat = gSNβ = 3 Nβ), which suggests the presence of significant axial magnetic anisotropy. Here, where the ground ±3/2 and excited ±1/2 Kramers doublets for D < 0, or the opposite for D > 0, are well separated due to sizeable ZFS effects operating on the quartet ground state, the magnetization for the ground Kramers doublet is the only one to be considered. Additionally, as observed in other highly anisotropic complexes reported [65], the isofield magnetization curves at low temperatures (2–10 K) are nearly superimposable for Co_4%@1 and Co_8%@1. Thus, M(H/T) curves were not useful in accurately determining the ZFS parameters but allowed us to conclude that the ZFS is quite large.

In order to quantify the relevant parameters, the experimental data for χ_MT vs. T and M vs. H of Co_4%@1 and Co_8%@1 were simultaneously simulated using the PHI program [66] through the spin Hamiltonian considering ZFS parameters (Equation (1)) for an isolated S = 3/2 state (solid line in Figure 3, Figure 4, Figures S7 and S8 and Tables S3 and S4), with axial (D) and rhombic (E) ZFS parameters. As the experimental measurements were performed on finely crushed polycrystalline samples, $g_x$ and $g_y$ components of the g-factor were considered indistinguishable. Thus, only the parallel and perpendicular components were retained ($g_x$ = $g_y$ = $g_{\perp }$ and $g_z=g_{\parallel }$). The best-fit values of the parameters ${g_{\parallel },g}_{\perp }, D, E, E/D$ were 2.24, 2.28, −55.25 cm^–1, 3.06 cm^–1, and 0.055 (Co_4%@1) and 2.41, 2.35, −50.80 cm^–1, 11.94 cm^–1, and 0.235 (Co_8%@1) cm^–1, which are in agreement with the literature values [37,65].

3.3.2. Dynamic (ac) Magnetic Properties

Alternating current (ac) magnetic measurements of Co_4%@1 and Co_8%@1 were carried out in the temperature range from 2 to 8 K to investigate their magnetization dynamics. No frequency dependence was observed for the ac components in the absence of a magnetic field (H_dc), a feature typically attributed to fast relaxation via quantum tunneling of magnetization. However, when different external static fields (1 and 2.5 kG) were applied, frequency-dependent in-phase (χ_M′) and out-of-phase (χ_M′′) signals were observed (Figure 5 and Figures S9–S11), which are consistent with a slow magnetic relaxation process typical of high-spin cobalt(II) SIMs.

Cole–Cole plots for Co_4%@1 and Co_8%@1 at 2–8 K under different applied static fields (1.0–2.5 kG) (Figure 6 and Figure S12) showed almost perfect semicircles, which were fitted using the Debye model (solid lines in Figure 6 and Figure S12). The obtained low values of the α parameters at different temperatures and dc magnetic fields [α = 0.099 (8.0 K)–0.121 (2.0 K) and 0.112 (8.0 K)–0.187 (2.0 K) for Co_4%@1 and Co_8%@1 under applied static field of 1000 G, respectively] (see Tables S3 and S4) support a single relaxation process. Moreover, increasing values of α with decreasing temperature reveal the appearance of intermolecular interactions within both frameworks of Co_4%@1 and Co_8%@1. The semicircles in the Cole–Cole plots are slightly more flattened in Co_8%@1 than in Co_4%@1 (Figure 6 and Figure S12). Thus, slightly larger α parameters were obtained for Co_8%@1 (Table S4). However, the α parameters are well below the values for spin-glass behavior (α ≥ 0.4).

Aiming to quantify the magnetic relaxation times, a joint analysis of the χ_M′ and χ_M′′ vs. frequency (υ) data (Figure 5 and Figure S9) was performed using the generalized Debye model. Two different approaches are recommended to represent these times depending on the nature of the dominant relaxation mechanism involved. After qualitative scrutiny, we established that lnτ vs. lnT plots (Figure 7 and Figure S13) were the most appropriate for analyzing the predominant relaxation mechanisms, commonly associated with Raman or direct-like mechanisms (${\tau }^{-1}\propto T^n$). Ideally, such plots should show a linear dependency with slope n. When n approaches a value close to 2 or 8, the optical or acoustic phonon-assisted Raman mechanism dominates the fastest spin reversal, while a value close to 1 indicates that the direct mechanism plays the main role. Here, we would like to note that the ideal behavior is often far away from reality in reported examples in the literature. Consequently, the confluence of various relaxation mechanisms operating at different temperatures is a common outcome. However, we aimed to keep this analysis simple and avoid overparameterization, which would eventually lead to better fitting but lack physical significance. In our case, lnτ vs. lnT plots for Co_4%@1 and Co_8%@1 at the two studied fields evidenced a deviation from linearity. The least-squares fit of these data yielded n values, which indicate that Raman mechanisms assisted by optical and acoustic phonons rule the relaxation of the magnetization at low and high temperatures, respectively (

Table 1. Selected parameters from the least-squares fit of the ac magnetic data of Co_4%@1 and Co_8%@1 ^a.

	Hdc (kOe)	C1 b (s−1 K−n)	n1 b	C2 b (s−1 K−n)	n2 b
Co4%@1	1.0	210 ± 11	2.52 ± 0.05	0.0018 ± 0.0007	8.84 ± 0.18
	2.5	153 ± 13	2.04 ± 0.09	0.024 ± 0.007	7.49 ± 0.15
Co8%@1	1.0	961 ± 30	1.94 ± 0.03	0.0037 ± 0.0013	8.58 ± 0.17
	2.5	575 ± 16	1.80 ± 0.03	0.017 ± 0.003	7.77 ± 0.08

^a The fits correspond to the double Raman relaxation model ${\tau }^{-1}={\mathrm{C}}_1{T}^{{\mathrm{n}}_1}+{\mathrm{C}}_2{T}^{{\mathrm{n}}_2}$. ^b Coefficient and polynomial factor for the Raman process.

).

4. Conclusions

In this work, we applied a partial transmetalation post-synthetic strategy to achieve two new MOFs that could not be obtained by direct synthesis. Through the controlled exchange of Zn(II) with Co(II) ions in the pristine diamagnetic Zn-MOF, two MOFs with 4% and 8% Co(II) content were obtained, exhibiting SIM behavior. This underexplored strategy holds great potential to enrich the elemental composition of an MOF in a controlled manner, thereby achieving a higher degree of complexity that leads to new physicochemical properties. In particular, it serves as an excellent tool for the preparation of molecular-based magnetic materials.