Fei-Yan
Yi
a,
Ying
Wang
a,
Jian-Ping
Li
a,
Dai
Wu
a,
Ya-Qian
Lan
*b and
Zhong-Ming
Sun
*a
aState Key Laboratory of Rare Earth Resource Utilization, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, 5625 Renmin Street, Changchun, Jilin 130022, P. R. China. E-mail: szm@ciac.ac.cn; Web: http://zhongmingsun.weebly.com
bSchool of Chemistry and Materials Science, Nanjing Normal University, Nanjing 210023, P. R. China. E-mail: yqlan@njnu.edu.cn
First published on 27th November 2014
In this work, a highly stable MOF luminescent switch {Cd3(L)(bipy)2·4DMA}n (1) has been successfully constructed, which exhibits clear fluorescence enhancement and “turn-off” quenching responses for benzene and nitrobenzene vapors, respectively, with high selectivity and sensitivity, as well as being fully reusable. Remarkably, the porous MOF (1) remains intact in aqueous solution over an extensive pH range of 2–13. This MOF sensor realizes fast detection for benzene vapor with a response time of less than one minute and ∼8-fold fluorescence enhancement. Furthermore, it as a porous multifunctional MOF also shows fully reversible adsorption behaviour for benzene vapor at room temperature. Thus the MOF material will be a promising luminescent sensor and adsorbent material for benzene vapor with important practical applications from environmental and health points of view.
Conceptual insightsFluorescence chemosensors as an expedient detection technique have been widely used for the detection of common volatile organic compounds (VOCs). They have many advantages over other sophisticated electrochemical devices, such as high signal output, simple detection, low expense, and reliability. Metal-organic framework (MOF) materials are promising luminescent sensors as their structures and functions are systematically and predictably designable and readily modulated. The reported MOF sensors have also verified their effectiveness, but most of those explored thus far show only a single “turn-off” or “turn-on” luminescence response for solvent solution. These are not efficient enough for detecting trace solvent vapors. In this paper we successfully synthesized an ultrastable MOF sensor, which realizes excellent ‘on–off’ switch-functions for the probing of benzene and nitrobenzene vapor. We investigate its various parameters to assess the practicality of the sensor, such as selectivity, sensitivity, response rate, repeatability and stability. This MOF-based luminescent switch opens up a new sensor platform with unprecedented practical applications. |
1a (30 mg) was placed into a glass tube (5 mL), then put under different solvent vapors for 24 h, including benzene, benzene-d6 (C6D6), toluene, 1,2-dichlorobenzene, 4-chlorotoluene, bromobenzene, nitrobenzene, nitromethane, methanol, ethanol, CH2Cl2, CCl4, 2-propanol, phenylmethanol and H2O. Subsequently the sample tube was taken out and quickly sealed, then the emission spectra were recorded.
The 1a sample used as in situ time-dependent photoluminescence sensor for solvent vapors was prepared as follows. A ground sample of 1a (100 mg) was pressed into a firm sheet, which was attached to a quartz slide and placed into a cuvette about 1.2 cm from the bottom. The cuvette containing the sensor slide was positioned on the solid sample holder of a Hitachi F-7000 spectrophotometer. The solid state emission spectra of firm sheet sample were recorded as an initial standard, then ∼120 μL benzene (or benzene-d6, nitrobenzene) solution was carefully added into the cuvette. The volume of cuvette used was 2 cm3 with a length and width of 1 cm and a height of 2 cm. So the concentration of the targeted solvent vapor in the cuvette was about 52.6 mg cm−3 (0.67 mmol cm−3).
Single-crystal X-ray diffraction analysis reveals that 1 crystallizes in a monoclinic C2/c space group (Table S1†).‡ In 1, two Cd(1) situated at the two ends and one Cd(2) at the center form a trinuclear secondary building unit (SBU) (CdII3) (Fig. 1a). In this SBU, Cd(1) is sevenfold-coordinated by two N atoms of one 2,2′-bipy ligand and five carboxylate oxygen atoms from three L6− ligands in a distorted {CdN2O5} pentagonal bipyramid geometry. The Cd(2) ion lies at an inversion centre and thus has 1/2 occupancy, adopting the typical {CdO6} octahedral coordination mode with six carboxylate oxygen atoms of anti–anti coordination conformations from four L6− ligands (Fig. S1†). The bond lengths of Cd–O and Cd–N in the range 2.166(7)–2.587(6) Å and 2.330(9)–2.392(11) Å, respectively, are comparable to reported values.12 Each unique L6− ligand with six carboxylate arms links ten Cd atoms by four μ2-η1,η2 carboxylate groups and two μ2-η1,η1 ones. Each SBU is connected to four adjacent units via four L6− ligands to further form a 2D double layer (Fig. S2†), which stack together to give the overall 3D structure with one-dimensional (1D) open channels having approximates sizes of 16.3 × 11.7 Å2 and 11.5 × 10.9 Å2 along the c-axis and 5.1 × 8.5 Å2 along the b-axis (measured between opposite atoms without taking van der Waals radii of the atoms into account) (Fig. 1b and c). The free DMA guest molecules reside in the channels (Fig. S3†). The overall structure can be simplified into a (4,4)-connected uninodal sql double layer (Fig. S4†) analyzed by the freely available computer program TOPOS.13 The solvent-accessible volume was calculated to be 43.3% by PLATON.14
Fig. 1 (a) CdII3 SBU. (b) View of the 3D porous framework along the c-axis. (c) The face-filling representation of 1. Color mode: Cd, green; C, grey; N, blue; O, red. |
The TGA result of activated 1 (see the ESI and Fig. S5†) indicates 1 may completely release its free solvent molecules to form guest-free 1 (referred to as 1a hereafter). The resulting 1a is thermally stable up to 265 °C. Adsorption and desorption isotherms of N2 (Fig. 2a) for 1a exhibit a single step type I curve with Brunauer–Emmett–Teller (BET) and Langmuir surface areas of 475.0 and 698.6 m2 g−1, respectively, as well as a total pore volume of 0.25 cm3 g−1. The pore size distribution (Fig. S6†) is around 8.6–9.3 Å, matching well with the crystal structure model. The adsorption isotherms of H2 at 77 and 87 K, and of CO2 at 273 and 298 K, were also measured up to 1 atm (Fig. 2a). They show smooth type-I isotherms with uptakes of 129.8 cm3 g−1 at 77 K and 96.0 cm3 g−1 at 87 K for H2, 50.9 cm3 g−1 at 273 K and 32.0 cm3 g−1 at 298 K for CO2. All of the above results demonstrate the robustness and the permanent microporous nature of 1a. More interesting is that 1 can be regenerated by soaking 1a in DMA. Such release and capture for solvent molecules are desirable characteristics, which prompted us to examine its ability as a potential sensory material and adsorbent for solvents. In addition, 1a is insoluble in water and common organic solvents, and it can still remain intact in the air for 3 months and in aqueous solution for 7 days (Fig. 2c). Each ground sample of 1a (30 mg) was immersed into aqueous solutions with different pH values from 1 to 13 for 24 h, then separated by simple centrifugation, washed three times with water (3 mL), then dried at 50 °C for 12 h. Their PXRD results (Fig. 2d) reveal that the characteristic peaks maintain nearly the same positions as those of as-synthesized 1a over a wide pH range from 2 to 13, suggesting that 1a is chemically stable with strong acid and base resistance, which is prerequisite for many important industrial applications. Meanwhile, the PXRD patterns and FTIR spectra also further illustrate its good water/solvent stability after solvent treatment (Fig. S7–S10†). Such high water stability and chemical stability demonstrate that 1a is ultrastable. Adding its thermal stability and porous characteristics lays a firm foundation for the further practical study of this material in chemical sensing applications.
As shown in Fig. S11,†1a exhibits a broad luminescence emission centered at ∼427 nm upon excitation at 324 nm. Compared to those of the neutral ligands H6L (three bands: 382, 401 and 423 nm) and bipy (two shoulder peaks: 392 and 411 nm), the emission band of 1a covers the whole range of H6L and bipy ligands, concomitant with a red shift of ≈16–26 nm, hence it can be assigned to ligand-to-ligand charge transition (LLCT), admixing with metal-to-ligand transition (MLCT) for such Cd2+ coordination complexes.6b,k,l The ground powder of 1a was put under water and organic solvent vapors (such as methanol, ethanol, CH2Cl2, CCl4, 2-propanol, phenylmethanol, benzene, toluene, bromobenzene, 1,2-dichlorobenzene, 4-chlorotoluene, nitrobenzene, 2,4-dinitrotoluene, and nitromethane) for 24 h (Fig. S12†). Notably, 1a shows significantly different luminescence responses (Fig. 3 and S13†) toward the vaporized analytes: an ∼8-fold fluorescence enhancement and a strong ‘turn-off’ fluorescence quenching effect occurred after exposure to benzene and nitro-explosives vapors, respectively. The results of the enhancing or quenching levels are in accordance with the ones in liquid media (Fig. S14†). More dramatically, 1a after exposure to benzene (1a/benzene) exhibits a hypsochromic shift of ∼67 nm (λmax = 381 nm, violet color) compared with 1a/nitro-aromatic explosives (λmax = 448 nm, blue color; Fig. S13c†). Such a luminescence enhancement effect, where the presence of the analyte triggers an increase in the luminescence intensity, is rare; in particular, high selectivity for benzene vapor with a colour shift in the emission wavelength is much rarer and practical. Interestingly, when benzene-d6 (C6D6) was employed as the target vapor, a clearly different fluorescent intensity from 1a/benzene was observed (Fig. S13d†), and an ∼4-times fluorescence enhancement was obtained. To further understand the energy levels of 1a in different solvents environment and host–guest interactions, a theoretical calculation using the Gaussian 09 package was employed based on an idealized non-periodic cluster model constructed according to the experimental data and the computational details are depicted in the ESI.† The calculation results are shown in Fig. 4, and indicate that the electron density of the highest occupied molecular orbital (HOMO) is dominated by p-orbitals of the carboxylate ligand, and the lowest unoccupied molecular orbital (LUMO) is mainly distributed on the N and C atoms in the bipy aromatic ligand. So the energy transfer from the carboxylate ligand to bipy ligand contributes directly to the luminescence of 1a. The corresponding calculated energy gap is ca. 2.75 eV (ca. 450 nm), which is far from the experimental fluorescence data (427 nm). That may be caused by the computational method or the selected ideal structural model of (1) a non-periodic basic trinuclear cluster, and (2) deleting all free solvent molecules, because the crystal data is actually periodic and it is difficult to completely remove all free solvent molecules in its channels although some preparatory work has been done. Such ligand-based emission in MOFs is more desirable, since the bandgaps of these materials can be altered by changing the degree of conjugation in the ligand, affecting the intensity and position of emission.15 In host–guest chemistry, exchanged solvent molecules can diffuse into the channels, leading to the different fluorescent responses by analyte–ligand interactions.
Fig. 4 Schematic drawings of the electronic structures of 1a in benzene and nitrobenzene environments. |
For 1a/benzene, the calculated energy gap increases from 2.75 eV to 3.17 eV, and the difference (0.42 eV) relative to 1a is very close to the observed blue-shift in the experiment of the luminescence (0.49 eV). Its LUMO is at a high-lying π* antibonding state with higher energy, so energy transfer from benzene to the LUMO of compound 1 occurs and leads to a fluorescence enhancement upon excitation. The significant shifting of emission λmax and fluorescence enhancement could be ascribed to the π⋯π stacking interaction between MOF host and benzene molecules, so that the energy levels of the ground state and excited state were changed (Fig. 4). In contrast, the excited electrons are transferred from the LUMO of 1 to nitrobenzene, thereby leading to a quenching effect for 1a/nitrobenzene. This mechanism is consistent with one previously proposed by other groups and us.9f,16 The UV-Vis adsorption spectra (Fig. S15†) of 1, ligands (H6L and bipy) and solvents (benzene and nitrobenzene) further verify and provide deeper understanding of the luminescent response. It is obvious that the adsorption peaks of 1 and bipy ligand at 220–322 nm almost cover the entire range of the absorption of solvents. Upon illumination, there is competition for excitation energy between the absorption of nitrobenzene and 1 because of the presence of the electron-withdrawing –NO2 group, whereas benzene molecules may have a similar excited energy level to 1a, which has been confirmed by experimental results. As shown in Fig. S16,† the fluorescent spectrum of pure benzene solution exhibits a main sharp emission maxima at ca. 328 nm and one very weak sloping band at ∼390 nm under normal 286 nm excitation. When under excitation at 314 nm, which is the excitation condition of 1a, benzene molecules can emit very weakly at λmax = 386 nm. Combined with the UV-Vis absorption data, it is most likely to lead to energy transfer from benzene to 1a or bipy ligands under excitation, which results in the increase of the luminescent intensity.11b
In order to test the sensitivity and response rate of 1a as a vapor sensor, an in situ solid-state luminescent sensor setup was designed and used (Fig. 5). The time-dependent enhancing and quenching efficiency shows that the response rate of 1a for benzene vapor is 56 s, and it exhibits ∼8 fold fluorescence enhancement in 2–3 min (Fig. S17†); for benzene-d6 vapor it is 30 s and reaches the maximum (∼4 times fluorescent intensity) in 2 min (Fig. S18†); and the fluorescence quenching for nitrobenzene reaches a maximum in 1 min with a quenching percentage of more than 90% (Fig. S19†). Such rapid, sensitive response to benzene vapor is still challenging. Repeatability for a sensory material is a very important parameter to assess the sensor practicability. After detection of benzene and nitrobenzene, samples of 1a/benzene and 1a/nitrobenzene can be recovered by simply heating at 80 °C under vacuum for 3 h, which can be evidenced by the PXRD patterns (Fig. S8†), FTIR spectra (Fig. S10†) and TGA curves (Fig. S20†). For 1a/benzene, its PXRD characteristic peaks match well with the ones of as-synthesized 1a, just the relative intensity shows a slight change (Fig. S8c†), which may be attributed to the benzene molecules filling into the MOF channels, giving rise to the high electronic density in this direction. The FTIR spectrum of 1a/benzene exhibits an obviously increasing infrared absorption peak around 676 cm−1 that belongs to the frame vibration absorption of benzene rings (Fig. S10k and l†) in the channels of 1a. The PXRD of 1a/nitrobenzene undergoes a small shift of ∼0.23° toward low angle and shows a few new peaks compared with 1a. So the interlayer distance along the (200) direction increases by 1.1 Å according to Bragg's equation. These changes evidence the structural change of 1a, caused by a large number of nitrobenzene molecules inserting in the channels of 1a as well as the hydrogen bonding and π⋯π interactions between the 1a host and nitrobenzene molecules. The IR spectrum for 1a/nitrobenzene further verifies (1) the presence of nitrobenzene molecules, where –NO2 characteristic νs and νas stretching vibrations (∼1519 cm−1 and ∼1343 cm−1) as labelled can be apparently observed (Fig. S10m and n†); (2) the stretches are associated with the inserted nitrobenzene molecules; (3) the framework of 1a/nitrobenzene keeps constant as 1a. The PXRD and IR results for 1a after heating treatment from 1a/benzene are in accordance with the ones of 1a-as-synthesized. The PXRD pattern for 1a regenerated from 1a/nitrobenzene also returns to be consistent with the initial one of 1a-as-synthesized. The –NO2 characteristic peaks decrease clearly in its IR spectrum. The changes during the adsorption and desorption processes suggest the “breathing effect” occurs in the flexible framework. Then a recovered 1a sample was reused in the next cycle. The results are presented in Fig. 4b and S17–S19.† It was observed that the response rate and sensitivity after three cycles were mainly maintained, and its PXRD patterns after treatment with benzene and nitrobenzene match well with the simulated pattern generated from the result of single-crystal diffraction data and as-synthesized product (Fig. S8†), confirming its good stability and reusability. The foregoing two aspects of rapid response and repeatability are of great significance for practical use of the sensor. One interesting feature can be easily observed during the process of in situ luminescent test: the emission wavelength of 1a/benzene is gradually blue-shifted as time increases. To deeply understand the influence of retention time for 1a in benzene vapor on its luminescence, a series of detailed studies were carried out, where the PL spectra of 1a in excess benzene vapor for different time intervals were recorded and are shown in Fig. S21.† These results show that the blue-shift is very rapid at the beginning and is complete after about 6 h, which is exactly identical with the foregoing results in Fig. 3 and 5.
As depicted in the experimental section, all in situ fluorescent tests were observed based on a concentration of 52.6 mg cm−3 for benzene vapor (∼120 μL). The fluorescent intensity change is similar to that with higher benzene concentration as shown in Fig. 3, where the sample was placed in excess benzene vapor. But it is highly possible that the luminescence enhancement is dependent on the vapor concentration. In order to further clarify the relationship, in situ fluorescent experiment was carried out in benzene vapor of a low concentration (8.8 mg cm−3; Fig. S22†). The observation indicates 1a/benzene still exhibits a clear enhancement effect, but only ∼3-fold luminescence enhancement was observed, which is far less than the ∼8-fold enhancement in the vapor concentration of 52.6 mg cm−3. This study demonstrates not only that the enhancement is significantly dependent on the concentration of benzene vapor, but also suggests that 1a may be a potential adsorbent material for benzene.
At the same time, one more important feature is also found that the firm sheet sample used for in situ solid-state luminescent testing expanded significantly, even breaking into a lot of small pieces after exposure to benzene vapor, yet not after exposure to nitrobenzene (Fig. S23†). This result may be caused by a large amount of adsorption for benzene molecules, so 1a is a potential MOF adsorbent for benzene vapor. As expected, the total weight of the solid sheet for 1a after treatment with benzene vapor increases by about 55–80% (repeating three times) of the original weight. The TGA curve of 1a after immersing in benzene (Fig. S20†) shows that a weight loss of ∼20% (200 mg g−1) from 50 to 250 °C is observed, which is equivalent to approximately 6 molecules per formula unit stably trapped inside the channels. But in fact, the weight loss of benzene adsorbed on the surface began at room temperature, so the amount of benzene molecules trapped by the framework of 1a is much greater than 20%. Furthermore, the sorption isotherm of benzene (Fig. 2b) by 1a has been measured and shows a rapid increase in the low-pressure range (P/P0 < 0.1), and then slowly approaches saturation with the amount adsorbed of 81.6 cm3 g−1 at P/P0 = 0.99, confirming the good adsorption properties of 1a. The desorption curve shows pronounced hysteresis and incomplete desorption, and does not retrace the adsorption isotherm, which may be attributed to the pore expansion in 1a when benzene molecules are adsorbed into the pores. Such a large adsorbed amount of benzene vapor into the channels may be another reason for the luminescence enhancement. But the amount for each of the other solvent vapors is difficult to measure, so the correlation between the adsorbed amounts of vapor molecules and luminescence intensity is still not clear at this moment. Similarly, the sample of 1a after thermal treatment does not affect the next adsorption. In other words, highly luminescent 1a will be a potential adsorbent for the removal of benzene molecules based on its high porosity, robust stability and reusability, which is needed as an environmentally friendly and effective method.
Footnotes |
† Electronic supplementary information (ESI) available: Materials and methods, crystal data, structural information, PXRD, FTIR, TGA, fluorescence measurements, UV-Vis and additional figures. CCDC 1017377. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4mh00210e |
‡ Crystal data for 1: C88H92Cd3N8O23, M = 1966.90, monoclinic, space group C2/c, a = 31.136(9) Å, b = 20.048(6) Å, c = 15.849(4) Å, V = 9390(4) Å3, Z = 4, μ = 0.746 mm−1, Dc = 1.391 Mg m−3, F(000) = 4016, 8371 unique (Rint = 0.0792), R1 = 0.0841, wR2 = 0.2592 (I > 2σ(I)), GOF = 1.011. Max./min. residual electron density 1.653 and −2.053 e Å−3. A total of 10440 data were measured in the range 1.23 < θ < 25.17°. |
This journal is © The Royal Society of Chemistry 2015 |