Metal-organic frameworks (MOFs) with intrinsic porosity and tailorable surface chemistry are an emerging class of proton-conducting materials for clean-energy related applications [1,2]. In order to improve proton conductivity of the material, the proton carrier concentration and mobility must be increased. To this end, great efforts have been made in developing water-mediated proton conducting MOFs because water can offer conjugated acid-base pair of H₃O⁺-H₂O and contributes to the construction of efficient proton-conducting pathway [3–7]. While the concentration of proton carriers can be increased by introducing acidic groups in the framework and/or acidic counterions and guests [8–12]. Besides water, NH₃ is also considered as a promising medium for proton conduction because of its similarity to water [13]. Recent reports have shown that exposing the material in NH₃-H₂O vapor can significantly increase the proton conductivity by 1–2 orders of magnitude [14–19]. However, it is unclear whether the coexistence of conjugate NH₄⁺-NH₃ and H₃O⁺-H₂O pairs can further enhance the proton conductivity. To answer this question, MOFs with good ability to adsorb ammonia and possessing structural stability in basic media are highly desired.

Metal phosphonates are known to exhibit high thermal and chemical stability because each phosphonate group can afford three oxygen atoms to bind metal ions [20]. The water stability of trivalent and tetravalent metal phosphonates is higher compared to monovalent and divalent metal compounds [21]. By selecting suitable metal ions and phosphonate ligands, porous metal phosphonates containing P-OH acidic groups can be designed and synthesized for ammonia adsorption and proton conduction [22–26]. Herein, we report a mixed-ligand dysprosium framework Dy₂(amp₂H₂)₂(mal)(H₂O)₂·5H₂O (MDAF-6), where amp₂H₄ is a pre-photodimerized 9-anthrylmethylphosphonic acid and malH₂ is malonic acid. The amp₂H₄ was chosen because it contains a large and flexible dianthracene group, which facilitates the formation of a 3D framework structure [27–29]. Dy^Ⅲ ion was chosen because its trivalent state can improve the framework stability and its unique magnetic properties lead to multifunctional materials [30–36]. Indeed, MDAF-6 shows a 3D open framework structure in which the uncoordinated P-OH group dangles over the channel surface. It exhibits excellent NH₃ adsorption capacity due to the acid-base reaction between NH₃ and P-OH groups. The resulted material, MDAF-6-NH₃, shows five orders of magnitude enhancement in proton conductivity compared to MDAF-6 in saturated NH₃-H₂O vapor, indicating the importance of the coexistence of NH₄⁺-NH₃ and H₃O⁺-H₂O conjugate acid-base pairs for proton conduction (Scheme 1). The magnetic properties of MDAF-6 were also studied.

Compound MDAF-6 was obtained as pale-yellow rhombus crystals after solvothermal reaction of DyCl₃·6H₂O, amp₂H₄ and malonic acid in CH₃OH/H₂O at 90 ℃ for 48 h, and was characterized by PXRD, TG and IR measurements (Figs. S1–S3 in Supporting information). Single crystal structural analysis revealed that it crystallizes in the orthorhombic system, polar space group Pmn2₁ (No. 31) (Table S1 in Supporting information). The asymmetric unit contains two kinds of Dy atoms (each with half occupancy), one amp₂H₂²⁻ ligand, half an occupied malonate, one coordinated and 2.5 lattice water molecules (Fig. S4 in Supporting information). The lattice water molecules were treated with PLATON/SQUEEZE program because of heavy disorder, and the number was determined by elemental and thermal analyses (Fig. S2 in Supporting information). Both Dy atoms reside on the mirror plane, and each is seven-coordinated by four O atoms from four amp₂H₂²⁻ ligands (O1, O1A, O5B and O5C for Dy1; O2, O2A, O4D and O4E for Dy2), two O atoms from the chelated malonate anion (O7 and O9 for Dy1; O7 and O8 for Dy2), and one from water molecule (O1W for Dy1; O2W for Dy2) (Fig. 1a). The Dy-O bond lengths and O-Dy-O angles are 2.218(9)‒2.465(16) Å and 69.2(2)°‒156.4(4)° for Dy1, and 2.232(9)‒2.749(9) Å and 50.5(3)°‒167.0(4)° for Dy2 (Table S2 in Supporting information). The {DyO₇} core has a distorted capped trigonal prism geometry (CShM = 0.755 for Dy1 and 1.434 for Dy2, Table S3 in Supporting information) [37]. The adjacent Dy1 and Dy2 atoms are connected alternatively by triple bridges of two O-P-O and one μ-O and double bridges of O-P-O units (Dy1^…Dy2 distance: 4.558, 5.389 Å), forming an infinite polar chain running along the b-axis, where the disordered malonate ligands locate on the same side (Fig. 1b). Each amp₂H₂²⁻ serves as a tetradentate ligand binding to four Dy atoms, whereas each mal²⁻ also acts as a tetradentate ligand chelating and bridging the Dy1 and Dy2 atoms (Fig. S5 in Supporting information). The neighboring chains are cross-linked by the amp₂H₂²⁻ ligands forming a 3D framework structure (Fig. 1c). A 1D channel is generated along the b-axis with the window dimension of ca. 7.0 × 6.2 Ų (van der Waals radii not accounted). It is noted that one of the two protonated phosphonate oxygen atom (O6) forms intrachain H-bond with the malonate oxygen atom O8, while the other (O3) points towards the channel center, and should form H-bond with lattice water molecule. Thus, the channel wall is hydrophobic on the dianthracene side, but hydrophilic on the chain side with pendent P-OH, malonate oxygen atom, and coordination water molecules.

To examine the stability of MDAF-6, the crystalline sample was immersed in water with different pH (1–14) for 24 h at room temperature. The PXRD measurements confirmed the stability of the material at pH 2–12 (Fig. S6 in Supporting information). Thermal analysis revealed that MDAF-6 lost seven water molecules below 170 ℃ (obs. 7.2%, calcd. 7.7%). The weight loss above 170 ℃ is due to the decomposition of organic components and the collapse of the framework structure (Fig. S2).

We next investigated the adsorption/desorption performance of MDAF-6, activated under vacuum at 100 ℃ for 4 h (weight loss: obs. 7.8%, calcd. 7.7%), toward N₂, CO₂, NH₃ and H₂O gasses (Fig. 1d). The N₂ adsorption/desorption isotherm at 77 K is a Type Ⅱ isotherm with a small loading of 4.1 cm³/g at P/P₀ = 0.98, indicating a very weak interaction between nitrogen and pore walls. The adsorption was increased for CO₂ at 298 K with a maximum uptake of 23.6 cm³/g (1.6 mol/mol) at P/P₀ = 0.98. A quick and dramatically enhanced adsorption was found for NH₃ with uptakes of 92.0 cm³/g (6.1 mol/mol) at P/P₀ = 0.04 and 142.4 cm³/g (9.4 mol/mol, 6.35 mmol/g) at P/P₀ = 0.98 at 298 K, attributed to the favourable acid-base interaction between P-OH and NH₃. The capacity is higher than MIL-53 (4.40 mmol/g) but lower than NH₂−MIL-53 and MIL-100 (8.00 mmol/g) [38]. After desorption, there remained 75.0 cm³/g (4.9 mol/mol) of ammonia at P/P₀ = 0.001, which is higher than the expected value of 4.0 mol⁻¹ when all four P–OH groups per molecular formula interacted with NH₃ to form non-volatile PO₃²⁻-NH₄⁺ pairs. This result suggests that NH₃ may also occupy the vacancy left by the Dy atom after removing the coordination water. The formation of PO₃²⁻-NH₄⁺ pair breaks the intrachain H-bond between P-OH and malonate oxygen and weakens the O(P)-H bond, and thus should increase the proton concentration for conduction in the framework channel. The sample after adsorption/desorption of NH₃ at 298 K is named as MDAF-6-NH₃. Interestingly, the re-activated sample, after heating MDAF-6-NH₃ at 150 ℃ under vacuum for 10 min (weight loss: obs. 5.8%, calcd. 5.6%), showed similar capacity of NH₃ adsorption at P/P₀ = 1.0 (Fig. S7 in Supporting information).

The water adsorption isotherm of the fully dehydrated sample of MDAF-6 showed a quick uptake of water vapor, and the value reached 118.7 cm³/g (7.8 mol/mol) at P/P₀ = 0.08 and 226 cm³/g (15 mol/mol) at P/P₀ = 0.99 at 298 K (Fig. S8 in Supporting information). When desorbed, there were still about 100.3 cm³/g (7 mol/mol) H₂O in the structure at P/P₀ = 0.03, in agreement with the presence of two coordinated and five lattice water molecules. We also measured the water adsorption capacity of MDAF-6-NH₃ at 298 K and observed an uptake of 78.4 cm³/g (5.3 mol/mol) at P/P₀ = 0.96, which is much lower than that for MDAF-6 (14.5 mol/mol). This is reasonable because the channel space in MDAF-6-NH₃ is partially filled with the loaded NH₃ molecules. In addition, the hysteresis of water adsorption/desorption isotherm is more significant for MDAF-6-NH₃ than for MDAF-6, indicating a strong interaction between water and NH₄⁺/NH₃. Notably, the water adsorption capacity of MDAF-6 and MDAF-6-NH₃ drops significantly upon slight increase of temperature. At 308 K, the capacities became 179.8 cm³/g (11.8 mol/mol) for MDAF-6 at P/P₀ = 0.98 and 67.1 cm³/g (4.4 mol/mol) for MDAF-6-NH₃ at P/P₀ = 1.00, respectively (Fig. S9 in Supporting information).

The proton conductivity (σ) of MDAF-6 and MDAF-6-NH₃ was evaluated by impedance spectroscopy measurements using a pellet sample placed in a temperature and humidity-controlled chamber for 12 h. For MDAF-6, the conductivity at 298 K was 1.8 × 10⁻¹² S/cm at 40% RH and increased to 1.5 × 10⁻⁹ S/cm at 95% RH (Fig. 2a and Fig. S10 in Supporting information). The values are very low compared to other proton conductive metal phosphonates with protonated phosphonate groups [22]. From the structure of MDAF-6, we can see that hydrophobic dianthracene moieties are present in the channel which is unfavorable for the formation of continuous hydrogen bond network. Besides, half of the protonated P-OH groups participate in strong intrachain hydrogen bonds, which may 'freeze' the protons, prevent them from migration, and thus affect their conduction. The adsorption of ammonia lifted the freeze of protons due to the acid-base interaction between P-OH and NH₃ to form NH₄⁺ in MDAF-6-NH₃. Proton conductivity measurements at 298 K revealed a remarkably improved conductivity of 1.0 × 10⁻¹⁰ S/cm at 40% RH and 6.8 × 10⁻⁶ S/cm at 95% RH (Fig. 2a and Fig. S11 in Supporting information). The latter is three orders of magnitude higher than that for MDAF-6. This value is comparable to some metal-triphosphonates [22,39], but lower than a few 3D metal-tetraphosphonates (ca. 10⁻²–10⁻⁴ S/cm) [26,40,41]. Obviously, the NH₄⁺ ions formed in-situ promote the water-mediated proton conduction.

The temperature-dependent proton conductivities of MDAF-6 and MDAF-6-NH₃ were measured at 95% RH. The proton conductivities were almost unchanged for both compounds when the temperature went up to 35 ℃ and further reduced above 35 ℃ (Fig. S12 in Supporting information), in consistence with the partial release of water molecules at this temperature. Based on the data in the temperature decreasing process, the activation energies (E_a) were estimated to be 0.35 eV for MDAF-6 and 0.61 eV for MDAF-6-NH₃ (Fig. 2b). The larger E_a (> 0.6 eV) suggests that NH₄⁺ may migrate directly as a proton attached to a vehicle in MDAF-6-NH₃.

The above results showed that the water-mediated proton productivity of MDAF-6 was significantly enhanced after the incorporation of NH₄⁺. In this case, there exist two acid-base pairs, e.g., H₃O⁺-H₂O and NH₄⁺-H₂O, which participate in the proton conduction pathway. We postulate that the presence of conjugate acid-base pair of NH₄⁺-NH₃ would facilitate the proton mobility.

To confirm this, we first evaluated the proton conductivity of MDAF-6-NH₃ under a dry NH₃ gas atmosphere at 100 kPa. The sample pellet was dried at 100 ℃ under vacuum for activation using a hot stage with electrical probes and then filled with dry NH₃ gas after cooling to room temperature. After being exposed to NH₃ gas for about 7 h, the gas adsorption of the sample pellet reached equilibrium and exhibited a conductivity of 3.4 × 10⁻⁹ S/cm at 304 K (Fig. S13 in Supporting information). This value is higher than that obtained at 298 K and 40% RH (1.0 × 10⁻¹⁰ S/cm), but much lower than that obtained at 298 K and 95% RH. The result indicates that the filling of NH₃ gas is not sufficient to form a continuous hydrogen bond network for efficient proton conduction. We then exposed the same pellet of MDAF-6-NH₃ to saturated NH₃-H₂O vapor. Interestingly, the proton conductivity drastically increased along with prolonged time and reached a maximum value of ca. 4.2 × 10⁻⁴ S/cm after 8.5 h (Figs. 2c and d, Fig. S14 in Supporting information). Controlled experiments were carried out under the same conditions for MDAF-6 in water or NH₃-H₂O vapor. As shown in Fig. 2c, the proton conductivity of MDAF-6 was ca. 10⁻⁹ S/cm after exposure to water or NH₃-H₂O vapor for 8.5 h, which is 5 orders of magnitude lower than that for MDAF-6-NH₃. The conductivity of MDAF-6 reached the equilibrium value of 2.1 × 10⁻⁸ S/cm after 32 h exposure to water vapor, and 6.9 × 10⁻⁵ S/cm after 172 h exposure to NH₃-H₂O vapor (Fig. S15 in Supporting information). Notably, the framework structure of MDAF-6 remained the same after thermal treatment, the NH₃ gas adsorption/desorption process, and exposure to saturated NH₃-H₂O vapor (Fig. S16 in Supporting information).

To understand the mechanism of enhanced proton conductivity of MDAF-6-NH₃ in NH₃-H₂O vapor, we measured the solid-state IR and ¹H MAS NMR spectra of the samples. Fig. S3 shows the IR spectra of MDAF-6 and MDAF-6-NH₃. Compared to MDAF-6 (3619, 3522 and 3441 cm⁻¹), the O-H stretching vibrations of MDAF-6-NH₃ are red-shifted to 3603, 3499 and 3418 cm⁻¹, indicating the weakening of the O-H stretching vibration. In addition, there appears a new broad peak at 3196 cm⁻¹, attributed to the N-H stretching vibration. The P-O stretching vibrations are very different for the two samples with additional strong peaks appearing at 1042 and 1001 cm⁻¹ for MDAF-6-NH₃. These results corroborate with the fact that the P-OH acidic group reacted with NH₃ base forming PO₃²⁻-NH₄⁺ pair in MDAF-6-NH₃.

After exposure to NH₃-H₂O vapor for 8.5 h, both MDAF-6 and MDAF-6-NH₃ show broad bands at 3680–3280 cm⁻¹, ascribed to the presence of extensive hydrogen bonding networks (Fig. S17 in Supporting information). Compared to MDAF-6, MDAF-6-NH₃ shows an enhanced stretching vibration at ca. 3435 cm⁻¹ and additional bands at 3207, 1402 and 1207 cm⁻¹. The latter three peaks are attributed to the N-H vibrations of NH₄⁺ and NH₃ [42,43]. The broad N-H stretching peak at 3207 cm⁻¹ may be related to the formation of hydrogen bonds between NH₃ and NH₄⁺ [44]. The observation of these N-H vibrations for MDAF-6-NH₃ but not for MDAF-6 indicates that MDAF-6-NH₃ adsorbs NH₃-H₂O vapor quickly and reaches equilibrium within 8.5 h. By contrast, MDAF-6 adsorbs NH₃-H₂O vapor slowly and needs much longer time (ca. 172 h) to reach equilibrium. This fact gives us an opportunity to distinguish the role of the NH₄⁺-NH₃ conjugate pair in proton conduction. For MDAF-6 exposed to NH₃-H₂O or pure water vapor, there was negligible or no NH₄⁺-NH₃ conjugate pair generated within 8.5 h, and their conductivities were extremely low (ca. 10⁻⁹ S/cm²). Therefore, the significantly enhanced proton conductivity of MDAF-6-NH₃ in saturated NH₃-H₂O vapor (ca. 10⁻⁴ S/cm²) has to contribute to the coexistence of the NH₄⁺-NH₃ and H₃O⁺-H₂O conjugate pairs.

We also measured the solid-state ¹H MAS NMR spectra of the above three samples. As shown in Fig. S18 (Supporting information), a resonance signal appears at 4.2 ppm for MDAF-6 exposed to H₂O and NH₃-H₂O vapor for 8.5 h, corresponding to physisorbed water moving freely in the pore system of the material [45]. For MDAF-6-NH₃ exposed to NH₃-H₂O vapor for 8.5 h, the signal is slightly shifted to 5 ppm, which could be related to the presence of NH₄⁺(NH₃) and H₂O interactions.

Based on the above results, we propose the mechanism of proton conduction of MDAF-6-NH₃ in NH₃-H₂O vapor as below: (1) The formation of PO₃²⁻-NH₄⁺ pair breaks the intrachain H-bond between P-OH and malonate oxygen and weakens the O(P)-H interaction, thus increasing the proton concentration for conduction; (2) The NH₄⁺ ion forms H-bonds with neutral NH₃ and transfers proton with low energy barrier [44]; (3) Saturated NH₃-H₂O atmosphere provides a continuous network of H-bonds for efficient proton conduction (Scheme 2).

In order to explore the possibility of using MDAF-6 as a magnetic proton conductor [46,47], we also investigated its magnetic properties. As shown in Fig. S19 (Supporting information), the χ_MT value of MDAF-6 (28.17 cm³ K/mol per Dy₂) at room temperature agrees well with the spin-only value of 28.34 cm³ K/mol for two isolated Dy^Ⅲ ions (⁶H_15/2, S = 5/2, L = 5, g_J = 4/3). The χ_MT decreases progressively upon cooling, attributed to the thermal depopulation of Dy^Ⅲ Stark sublevels and possible weak antiferromagnetic interactions. Field-dependent magnetization measured from 2 K to 10 K showed unsaturation up to 70 kOe (Fig. S19, inset), suggesting the presence of magnetic anisotropy and/or lower lying excited states. The alternating current (ac) susceptibility data of MDAF-6 revealed a frequency dependence of the out-of-phase (χ'') signals under zero or 1 kOe dc field (Fig. S20 in Supporting information), characteristic of single-molecule magnet (SMM) behavior. But no maximum was observed, thus excluding the possibility to derive the energy barrier of the material.

In summary, we report a new porous dysprosium phosphonate framework Dy₂(amp₂H₂)₂(mal)(H₂O)₂·5H₂O (MDAF-6) which shows a high adsorption capacity toward NH₃ due to the presence of P-OH acidic groups. Impressively, we observed a dramatic enhancement of proton conductivity after exposing MDAF-6-NH₃ in saturated NH₃-H₂O vapor, demonstrating the importance of coexistent conjugate acid-base pairs of H₃O⁺-H₂O and NH₄⁺-NH₃ in promoting proton conduction. In addition, MDAF-6 shows SMM behavior at low temperature. This work not only provides a new example of proton conductive magnetic materials, but also a new approach to boost the proton conduction of MOFs by introducing conjugated acid-base pairs of H₃O⁺-H₂O and NH₄⁺-NH₃ in NH₃-H₂O media.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by grant from the National Natural Science Foundation of China (No. 21731003).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.108320.