Introduction

Over the past two decades, interest in metal−organic frameworks (MOFs) has increased significantly not only because of their intriguing architectures, high crystallinity, exceptional porosity and diverse modularity, but also due to their promising applications in various fields, such as gas adsorption and separation, optical luminescence, catalysis, energy storage and sensing1,2,3,4. Although a significant number of MOFs have been synthesized and their physical properties have been examined, MOFs are still quite new materials. Hence, the design and synthesis of different kinds of MOFs is necessary to gain more knowledge about their structural diversity and investigate their various properties. While MOF nodes can be composed of single metal ions, they can also be made up of discrete metal-containing clusters, so called secondary building units (SBUs). These metal-cluster SBUs offer an opportunity to design and synthesize highly connected, non-interpenetrating networks with enhanced framework stability and porosity. Among metal cluster SBUs, the Zn based clusters, such as di-, tri-, tetra- and pentanuclear zinc carboxylate clusters are particular useful to build porous networks, since they have a richer variety of size and geometry that allow for more elaborate structural design5. Indeed, a plenty of MOFs have been created by assembling Zn based clusters and organic ligands, however, those constructed by two different types of zinc carboxylate clusters, which may further facilitate the structural diversity of Zn-MOFs, are still rare6,7.

On the other hand, metal ion exchange is an emerging synthetic route for modifying the secondary building units of MOFs without changing their framework topology. This approach not only can improve the properties of MOF materials, but also allow the preparation of isomorphous MOFs in a single crystal-to-single crystal fashion that cannot be obtained through conventional synthetic routes. Cu2+ ion, for example is more likely to replace Zn2+ in MOFs. So far, such cation exchanges usually occur at single zinc nodes or paddlewheel zinc carboxylate units8,9,10, those that take place at zinc cluster SBUs are less known. In this work, we demonstrate the replacement of Zn2+ by Cu2+ at the tri- and tetranuclear zinc clusters in a nanocage-based MOF, resulting in the formation of a Cu analogue with enhanced gas adsorption properties.

Multi-colour emission materials (especially white light) have received increasing attention because they have shown great promise in a variety of applications, from displays, solar cells, to light-emitting diodes. Recently, MOFs have been utilized to generate tunable colour and white light emission through doping appropriate amount of Eu3+ and/or Tb3+ ions in a single lattice framework composed of Ln3+ or non-lanthanide metal ions11,12,13,14,15,16. This approach still remains a great challenge owing to the difficulty of precisely controlling the ratio of different Ln3+ ions in one single framework. Another alternative approach to realize colour-tunable luminescence is to incorporate Ln3+ species in some microporous luminescence MOFs. However, there are some limitations of these host-gust systems in terms of the judicious selection of suitable host framework and adjusting the incorporation amounts of different Ln3+ ions17,18,19,20,21.

In the past decade, luminescent MOFs have emerged as promising candidates for the rapid, sensitive and accurate recognition of metal ions22,23,24,25,26,27. The recognition of metal ions plays a very important role in many aspects, including our life28,29. The Cu2+ and Al3+ ions, for instance, are necessary for maintenance of human metabolism. Nevertheless, high concentrations of Cu2+ and Al3+ can lead to many adverse health effects. Therefore, the design and synthesis of luminescence MOFs capable of sensing Cu2+ and Al3+ is very important30,31. Herein we report a novel Zn-cluster based MOF, [Zn21(BTC)11(µ3-OH)3(µ4-O)3(H2O)18]·21EtOH (1) (H3BTC = 1, 3, 5-benzenetricarboxylic acid) built from a triangular Zn cluster SBU, a tetrahedral Zn cluster SBU and a tritopic linker BTC3−.Tunable colour and white light emission can be achieved by varying the excitation wavelength and incorporating appropriate amount of Eu3+/Tb3+ in the pore of 1. In addition, compound 1 also exhibits a great potential as a luminescence sensing material for Cu2+ and Al3+ ions.

Results and Discussion

Synthesis and description of crystal structure

Colourless crystals of 1 were synthesized by the solvothermal reaction of Zn(NO3)2·6H2O,H3BTC and 4-cyanopyridine in a 1:1:2 molar ratio, in ethanol (10 ml) at 110 °C for three days. Single crystal X-ray diffraction studies reveal that 1 crystallizes in the trigonal space group R3. The asymmetric unit of 1 contains seven Zn2+ ions, 11/3 BTC3− ligands, one μ3-OH− anion, one μ4-O2− anion and six coordinated water molecules (Fig. S1). The structure contains two types of Zn clusters. One is the trinuclear cluster [Zn3(µ3-OH)(COO)5(H2O)3] (simplified as Zn3O) and the other is the tetranuclear cluster [Zn4(µ4-O)(COO)6(H2O)3] (simplified as Zn4O). In the Zn3O cluster, there is a µ3-OH group located at the centre of the cluster. Three Zn ions in Zn3O adopt different coordination geometries: Zn5 resides in a distorted tetrahedral geometry, whereas Zn6 and Zn7 adopt a square pyramidal and an octahedral geometry respectively (Fig. 1a). The tetranuclear cluster consists of two ZnO4 tetrahedra, a ZnO5 square pyramid and a ZnO6 octahedron sharing a central µ4-O atom (Fig. 1b). The Zn–O bond lengths and angles varied in the normal ranges of 1.885(9)–2.382(16) Å and 83.0(5)–176.6(5)°, respectively. The average Zn···Zn separation in the Zn3O cluster is 3.357 Å which is slightly larger than that in the Zn4O cluster (3.196 Å). The BTC3− ligands adopt four different linking modes, denoted as I (linking three Zn4O clusters), II (linking two Zn4O and a Zn3O clusters), III (linking two Zn3O and a Zn4O cluster) and VI (linking three Zn3O clusters) (Fig. 1c) to connect Zn3O and Zn4O clusters into a large polyhedral cage (Fig. 1d). This cage is composed of nine Zn3O and nine Zn4O cluster vertexes linked by twenty-four triangular BTC3− ligands and may enclosed a sphere of ca. 18.5 Å diameter.

Figure 1
figure 1

(a) The Zn3O SBU; (b) The Zn4O SBU; (c) The linking modes of BTC3− ligand; (d) View of the octahedral cage constructed by Zn3O and Zn4O cluster nodes and BTC3− linkers; (e) View of the octahedral cage by connecting Zn3O and Zn4O cluster nodes; (f) Two octahedral cages are connected by sharing two edge-fused triangles; (g) View of the one octahedral cage surrounded by six identical cages.

Full size image

A better insight into this cage can be achieved through connecting the Zn clusters which generates a slightly distorted octahedron whose faces are each composed of four small triangular faces. The size of the octahedral cage is ca. 30.7 × 21.5 × 21.5 Å (Fig. 1e). It has been known that the C3-symmetric ligand H3BTC is useful for the construction of Zn-BTC octahedral cages. However, the short spacer of H3BTC usually leads to small cages32,33,34,35. While increasing the lengths of C3-symmetric ligands can afford large cages36, this work demonstrates that the Zn-BTC octahedral cage can also be expanded with Zn3O and Zn4O cluster nodes. In 1, each octahedral cage serves as a 6-connected octahedral node and shares two edge-fused triangles of six faces with six surrounding octahedra (Fig. 1f), generating a complicated 3D microporous framework (Fig. 1g). Topologically, the Zn3O and Zn4O clusters can be considered as distorted square pyramidal and octahedral SBUs (5- and 6-connected nodes, respectively), and the BTC3− linker as a triangular unit (3-connected node) (Fig. S2a and b). Thus, the 3D framework of 1 can be viewed as a (3, 5, 6)-connected net (Fig. S2c and d).

To prove the phase purity of the bulk sample, PXRD analysis is performed. The peak positions of the simulated pattern closely match those of the experimental one, indicating phase purity of the as-synthesized sample (Fig. S3). Thermogravimetric analysis (TGA) of 1 shows a mass loss of ca. 25.2% from 30 to 400 °C, which is corresponding to the loss of lattice solvent molecules and the coordinated water molecules (calcd. 25.6%). Upon further heating the framework starts to decompose (Fig. S4).

Tunable luminescence and white light emission

Compounds with d10 metal centres and organic ligands are desirable candidates for luminescence-emitting materials. Hence luminescence excitation and emission spectra of 1 and H3BTC were investigated at room temperature (Fig. S5). Compound 1 and H3BTC exhibit emission bands at 422 nm (λex = 355 nm) and 430 nm (λex = 340 nm). By comparison with the free ligand, the emission of 1 is blue shifted by ca. 8 nm. Such behaviour could be due to the strong electrostatic interaction between the Zn2+ ion and BTC3−. The solid-state luminescence of 1 excited with various wavelengths was also investigated. As shown in Fig. 2a, as the excitation wavelength varies from 310 to 471 nm, the luminescence colour changes from light-blue to blue-green (Fig. 2b). This result gives us an opportunity to obtain white light emission by incorporating red and green emitting components such as Eu3+ and Tb3+ into the pore of 1.

Figure 2
figure 2

(a) The emission spectra of 1 under excitation at various wavelengths; (b) The CIE values of 1 at different excitation wavelengths; (c) The luminescence emission spectra of Eu3+/Tb3+-1 by varying the excitation wavelength; (d) The CIE values of Eu3+/Tb3+-1 excited at different wavelengths.

Full size image

In order to make Ln3+-incorporated complexes, compound 1 was immersed in turn in an ethanol solution of Eu3+ and an ethanol solution of Tb3+, then the solid was filtered and washed by ethanol and diethyl ether several times to remove any residual Eu3+ and Tb3+ ions on the surface. By adjusting the immersion time, the encapsulated amount of Eu3+ and Tb3+ can be optimized to achieve white light emission. The resultant Ln3+-incorporated complex, namely Eu3+/Tb3+-1 contains 0.251% of Eu3+ and 0.269% of Tb3+, as confirmed by ICP results. The solid-state emission spectrum of Eu3+/Tb3+-1 exhibits the characteristic emission peaks of Eu3+ (5D0 to 7FJ, J = 0–4) and Tb3+ (5D4 to 7FJ, J = 6–0) (Fig. 2c). Notably, the CIE coordinates of Eu3+/Tb3+-1 excited at 312 nm is (0.32, 0.33), which are very close to those for pure white light (0.333, 0.333), according to the 1931 CIE coordinate diagram. Meanwhile, the emission of Eu3+/Tb3+-1 under different excitation wavelengths was also investigated. When excited at 308 nm, the CIE index of Eu3+/Tb3+-1 is (0.38, 0.40), and it shows yellow light emission. As the excitation wavelength increases gradually, the main emission peaks of Eu3+ and Tb3+ gradually weaken. When excited at 315 nm, the CIE of Eu3+/Tb3+-1 is (0.27, 0.29), and it displays a blue-green light. As a result, the luminescence colour of Eu3+/Tb3+-1 at different excitation wavelengths changes from yellow to white, and eventually becomes blue-green (Fig. 2d).

Luminescence sensing for metal ions

The existence of a porous structure makes compound 1 a promising candidate for sensing and detecting metal ions. To investigate the luminescence quenching or enhancement behaviour of 1 by various metal ions, solid samples of 1 were immersed in ethanol solutions containing 0.03 M of M(NO3)n (M = Al3+, Ga3+, In3+, Li+, Mg2+, Cd2+, Ca2+, Gd3+, Zn2+, Co2+, Ag+, Ni2+, Mn2+, Cu2+, n = 1−3) for one hour and then ultrasonically agitated for 20 min to form a metal-ion-incorporated MOF suspension. The corresponding luminescence spectra are recorded and are compared in Fig. 3a. The emission spectra show that the luminescence intensity of Mn+-1 excited at 355 nm varies significantly depending on the identity of the metal ions. For example, Li+, Mg2+, Ca2+, In3+, Zn2+, Cd2+ and Gd3+ have only a slight effect on the luminescence intensity after incorporation into the pores, whereas the other metal ions have varying degrees of effects. Among them the Cu2+ ion has a significant quenching effect on the emission of 1. The descending order of the quenching efficiencies of the metal ions is as follows: Cu2+ > Mn2+ > Ni2+ > Ag+ > Co2+ > In3+ > Li+ > Mg2+ > Zn2+ > Cd2+ > Ca2+ > Gd3+. In contrast to Cu2+, the Al3+ and Ga3+ ions show significant enhancement on the emission intensity. Particularly in the presence of Al3+, the emission intensity is about three times than the metal-ion free 1 (Fig. 3b). These results clearly indicate that 1 shows a high selectivity towards Cu2+ and Al3+.

Figure 3
figure 3

The luminescence spectra (a) and intensity (b) of 1 after treatment with different metal ions.

Full size image

The relationship between the luminescence intensity and the concentration of Cu2+ has been investigated by measuring the emission spectra of 1 after immersion in solutions of various concentrations of Cu2+ ions (Fig. S6a). The results show that the luminescence intensity of Cu2+-incorporated complex is greatly dependent on the concentration of the metal ion. The luminescence intensity decreases quickly as the concentration of Cu2+ increases and it remains unchanged when the Cu2+ concentration is greater than 0.03 M (Fig. 4a). Unlike Cu2+ ion concentration, the immersion time seems to have no influence on the luminescence intensity. As shown in Fig. 4b, the luminescence intensity of 1 after being immersed in 0.03 M Cu2+ ethanol solution for less one minute decreases sharply and it is also observed that prolongation of immersion time up to 60 min does not cause any further decrease of the luminescence intensity (Fig. S6b). Furthermore, this selective detection of Cu2+ is not influenced by the existence of other metal ions such as Li+, Zn2+, Cd2+ and Gd3+ (Figs 4c and S6c). A good linear correlation between (I0−I)/I and the concentration of Cu2+ is observed with the Ksv value of 286.1 M−1 (Fig. S6d). The detection limit is calculated on the basis of 3σ/k to be 1.34 × 10−3 M.

Figure 4
figure 4

The luminescence intensity of 1 (a) after immersion in ethanol solutions of different concentrations of Cu2+ ions (b) after immersion in a 0.03 M ethanol solution of Cu2+ for various time periods and (c) after immersion in ethanol solutions with different metal ions.

Full size image

The possible mechanism of luminescence quenching by Cu2+ could involve the binding of Cu2+ through Lewis acid-base interaction, as suggested for the selective sensing of Cu2+ ion with microporous frameworks, such as [Cd2(PAM)2(dpe)2(H2O)2]·0.5(dpe)37 and {Mg(DHT)(DMF)2}38 (H2PAM = 4,4-methylenebis(3-hydroxy-2-naphthalene-carboxylic acid), dpe = 1,2-di(4-pyridyl)ethylene, DHT = 2,5-dihydroxyterephthalate). Such binding reduces the intraligand luminescent efficiency and results in the quenching effect37,38,39,40. In the case of Al3+ sensing, decomposition of 1 occurred due to hydrolysis of Al3+, which made the solution acidic. In 0.03 M Al3+ solution, 1 was partially dissolved, which released BTC3− in solution and thereby enhancing the ligand fluorescence41.

The luminescence quenching of Cu2+ may also result from partially exchange of the metal ions from Zn2+ to Cu2+ in the framework. To check the results, 1 was immersed in ethanol containing 0.03 M Cu2+ for varying periods of time. The filtered powder was washed thoroughly with ethanol until the filtration became colourless. The colour of crystals changes from colourless to green after exchange with Cu2+. The Cu2+-exchanged samples thus obtained were then subjected to ICP analysis. The results showed that with immersion time increasing from 1 min to 24 h, the Cu-exchange level on the framework gradually increased accompanied by a reduction of Zn content in the compound. As indicated in Table S1, approximately 50% of framework was replaced with Cu2+ ion within one day and nearly complete replacement of Cu2+ (96%) took place after two weeks. Surprisingly, the reversed ion exchange failed and so did the exchange with other transition metal ions like Co2+ and Ni2+. SEM images (Fig. S7) reveal that after Cu2+ exchange, the large crystals of 1 (ca. 300 μm) with a regular shape collapse into microcrystalline solid. However, the XRD pattern of the Cu2+-exchanged samples shows similar peaks to those of 1, suggesting that the framework structure remains intact after MOF exchange with Cu2+ ion (Fig. S3). Moreover, XPS measurements were also carried out to confirm the existence of Cu in the Cu2+-exchanged sample. Fig. S8 shows that the Cu 2p 3/2 and 1/2 spectra of Cu2+-exchanged sample were located at 934 and 954 eV, respectively, both of which suggest the presence of Cu2+ in the Cu2+-exchanged sample.

Gas adsorption properties of the Cu2+-exchanged compound

Adsorption experiments were carried out to investigate the porosity. The samples were degassed at 100 °C for 12 h under vacuum prior to gas adsorption/desorption measurements. The activated sample 1 shows no significant adsorption for N2 and CO2, presumably due to the pore collapse during sample activation. Interestingly, the adsorption capacity for either N2 or CO2 substantially increases by replacing Zn2+ with Cu2+ in the framework. The N2 adsorption of the Cu2+-exchanged samples with 50% (1a) and 96% (1b) exchange ratios at 77 K exhibits a type I isotherm, typical for materials that show permanent microporosity. The highest adsorbed amount of N2 is 174.8 cm3 g−1 for 1a and 324.6 cm3 g−1 for 1b, and the corresponding pore volumes are 0.245 and 0.485 cm3 g−1 respectively (Fig. 5a). The Langmuir and Brunauer-Emmett-Teller (BET) surface areas are 649.02 and 589.05 m2g−1 for 1a and 1299.33 and 1179.73 m2g−1 for 1b. The CO2 adsorption capacity increases from 38.5 cm3 g−1 for 1a to 136.1 cm3 g−1 for 1b. In both case, the amount of CO2 uptake decreases by 40.3% as the temperature increases from 273 to 298 K, indicating a typical physisorption behavior (Figs 5b and S9). These results demonstrate that the increase of Cu-exchange ratio dramatically enhance the adsorption capacity for N2 and CO2. The main reason for this may be due to the fact that the replacement of Zn2+ by Cu2+ enhances the framework robustness thereby improving the adsorption properties42.

Figure 5
figure 5

(a) N2 adsorption-desorption isotherms of 1, 1a and 1b at 77 K; (b) CO2 adsorption-desorption isotherms of 1, 1a and 1b at 273 K.

Full size image

Conclusion

In summary, a new cage-based MOF with two different types of Zn cluster SBUs has been synthesized and structurally characterized. This compound features a large octahedral cage constituted by nine Zn3O and nine Zn4O clusters and twenty-four triangular BTC3− ligands. Tunable luminescence and white light emission can be achieved by changing the excitation wavelength and by incorporation of Eu3+/Tb3+ ions into the compound. While other transition metal ions such as Mn2+, Co2+ and Ni2+ displayed relatively weak quenching effects, only Cu2+ and Al3+ ions showed significant changes in the emission spectra, which demonstrates that 1 could be regarded as a potential material for selective sensing of Cu2+ and Al3+ ions. In addition, the facile ion exchange with Cu2+ without loss of structural integrity as described herein provide an post-synthesis route to construct isomorphous Cu-MOF that cannot be obtained by direct synthesis.

Methods

Materials and instrumentation

All chemicals were purchased commercially and used as received. TGA was performed using a TGA/NETZSCH STA449C instrument heated from 30–800 °C (heating rate of 10 °C/min, nitrogen stream). IR spectrum using a KBr pellet was recorded on a Spectrum-One FT-IR spectrophotometer in the range 4000-400 cm−1. The powder X-ray diffraction (PXRD) patterns were recorded on crushed single crystals in the 2θ range 5–55° using Cu Kα radiation. ICP elemental analyses for the metal ions were performed with an ultima2 X-ray ICP optical emission spectrometer. Elemental analyses for C and H were measured with Elemental Vairo EL III Analyser. Luminescence spectra for the solid samples were recorded on an Edinburgh Analytical instrument FLS920. Luminescence spectra for the liquid samples were recorded on a HITACHI F-7000. Gas adsorption measurements were performed in an ASAP (Accelerated Surface Area and Porosimetry) 2020 System. SEM images were obtained using a Phenom G2 SEM microscope.

Preparation of compound 1

A mixture of Zn(NO3)2·6H2O (148.7 mg, 0.5 mmol), H3BTC (103.5 mg, 0.5 mmol) and 4-cyanopyridine (104.1 mg, 1.0 mmol) in ethanol (10 mL) was heated in a Teflon-lined stainless steel vessel (24 mL) at 110 °C for three days and then cooled to room temperature in two days. The resulting colourless crystals of 1 were obtained and washed several times with ethanol (yield 56% based on Zn). Elemental analysis calcd. (%) for 1 C47H66O37Zn7 (1680.58): C 33.56, H 3.93; found: C 33.29, H 3.87. IR (cm−1) (Fig. S9): 3433 s, 2977 vs, 1687 w, 1574 s, 1440 s, 1365 vw, 1260 w, 1197 vw, 1105 w, 1046 w, 926 w, 875 w, 829 w, 762 s, 730 s, 553 w, 469 w.

Preparation of Eu3+/Tb3+-1

The Ln3+-incorporated complex was prepared by first soaking a sample of 1 (35 mg) in an ethanol solution (3 mL) containing Tb(NO3)3·6H2O (20 mg) for two hours, afterwards in a Eu(NO3)3·6H2O (20 mg) ethanol solution (3 mL) for another two hours. Then the crystals were collected, washed thoroughly with ethanol and diethyl ether, and dried in air to afford Eu3+/Tb3+-1.

Immersion experiments of 1 with different metal ions

Compound 1 (30 mg) was immersed in 0.03 M solutions of M(NO3)n in ethanol at room temperature for one hour (M = Al3+, Ga3+, In3+, Li+, Mg2+, Cd2+, Ca2+, Gd3+, Zn2+, Co2+, Ag+, Ni2+, Mn2+, Cu2+, n = 1−3) and then ultrasonically agitated for 20 min to form a metal-ion-incorporated MOF suspension.

Single-crystal structure determination

Single-crystal X-ray diffraction data were collected on a Rigaku Diffractometer with a Mercury CCD area detector (Mo Kα; λ = 0.71073 Å) at room temperature. Empirical absorption corrections were applied to the data using the Crystal Clear program43. The structure was solved by direct methods using SHELXS-9744 and refined by full-matrix least-squares on F2 using SHELXL-2016 program45. Metal atoms were located from the E-maps, and other non-hydrogen atoms were located in successive difference Fourier syntheses. All non-hydrogen atoms were refined anisotropically. The organic hydrogen atoms were positioned geometrically. Since the position of the disorder water molecules could not be resolved from Fourier maps, PLATON/SQUEEZE46 was used to compensate the data for their contribution to the diffraction patterns. The SQUEEZE calculations showed a total solvent accessible area volume of 10178 Å3 in 1 and the residual electron density amounted to 1744 e per unit cell, corresponding to about seven ethanol molecules per asymmetric unit. The final formula was then calculated from the TGA result combined with elemental analysis data. Crystallographic data and other pertinent information for 1 are summarized in Table S2. Selected bond distances and angles are listed in Table S3. CCDC number for 1 is 1542054.