2.1. Synthesis and Characterization
The microporous MOF was prepared by a solvothermal reaction involving the coordination of two highly conjugated ligands, H2
bpdc and bpee, with the Zn(II) metal ion. As can be seen in the powder X-ray diffraction (PXRD) analysis (Figure 2
) the as-synthesized PXRD pattern (red line) showed high similarities with the simulation based on single crystal X-ray diffraction (SCXRD) database (black line), indicating the successful synthesis of the material. This is in agreement with the results published by Lan et al. [12
As described by Lan et al. [12
], the MOF structure contains roughly rectangular 1D channels in which dimethylformamide (DMF) solvent molecules are encapsulated. To activate this material, the DMF molecules need to be removed, releasing the channels (pores) and thus making it suitable for detection purposes. Our initial attempt to activate the material involved the treatment in CH3
OH (three days) and CH2
(four days) for solvent exchange, followed by a drying period under vacuum at room temperature. Following this protocol, the PXRD analysis revealed that the intended MOF was obtained (Figure 2
, blue line). However, our initial studies in gas detection showed a much lower sensitivity than reported, which we attributed to the material not being properly activated because the pores of the MOF had not been fully released from the DMF molecules, hindering the gas diffusion through the sensing material. Considering this, an alternative activation strategy to remove the uncoordinated DMF was tested, which involved the solvent exchange and a thermal drying procedure at 493 K under vacuum (30 mbar) for 5 h. In terms of PXRD analysis, Figure 2
presents the comparison for the MOF obtained from the first process of DMF removal (solvent exchange) and the second one (drying under vacuum at 493 K). The crystallinity and main structural features of the material were maintained after both processes, which correspond to the activation of the MOF. Apparently, the activation of the MOF by the removal of DMF molecules from the pores causes small structural changes, as reflected by the few alterations verified in the PXRD patterns. However, it is evident that the MOF material maintains considerable permanent porosity, allowing its potential activity in sensing.
The incomplete elimination of DMF from the MOF after the solvent exchange was confirmed by FTIR spectroscopy. The presence of a characteristic C=O stretching peak at ~1670 cm−1
proved the existence of DMF molecules occluded into the material (Figure 3
]. Upon drying the MOF under vacuum at 493 K, DMF in the MOF was efficiently removed, as supported by the disappearance of the DMF C=O stretching peak from the FTIR spectrum of the desolvated sample (Figure 3
). This procedure for DMF removal (pore activation) present a significant advantage in comparison with the reported one, as it took less than three days to obtain the MOF, instead of the seven days reported by Lan et al. [12
]. The material prepared using the second process was the one used for the gas sensing tests due to its better performance in terms of sensitivity and gas diffusion.
SEM images of MOF showed the material in the form of microneedles alongside bigger aggregates with no particular shape (Figure 4
). Such an arrangement leads to a high surface-to-volume ratio due to the small diameter of the microneedles, favoring the gas-sensing capabilities of our material due to the increase in exposed surface area. This adds to the fact that the MOF is highly porous owing to its internal channels [12
], hence allowing easy diffusion of gas molecules through the internal structures and facilitating their accessibility to the material active sites.
2.2. Soft-Imprinted MOF Films
Soft-imprinting of MOF on CA-coated quartz substrates (CA thickness: 200 nm) with an applied pressure of 2, 4, and 6 bar resulted in MOF particles partially immersed into the CA film, as can be seen in atomic force microscope (AFM) images (Figure 5
). In general, MOF crystals were dispersed on the surface and embedded but not covered by the CA, clearly protruding from the film. According to the images, individual microneedles were found, lying flat on the surface, along with bigger aggregates. In the analyzed regions, the average height of the microneedles from the CA surface was found to be around 5 nm, while that of the aggregates was approximately 95 nm. Further characterization of the films was performed by analyzing the roughness of the AFM images. Root mean square roughness for all samples was similar and in the order of 5 nm, indicating on one hand that the three different pressures applied to imprint the MOF led to films with similar surface characteristics. On the other hand, the roughness of the films points at the inhomogeneity of their surfaces, which can be beneficial for gas-sensing purposes due to a high surface area to volume ratio and improved access of gas molecules towards the active sites of the films, i.e., MOF particles. For comparison, AFM images of a pure CA film were also analyzed, revealing a smooth and even surface prior to the soft-imprinting of MOF (Figure S1
). The CA layer was found to efficently retain the MOF particles, providing mechanical stability to the sensor. By using a thin CA film, we aimed to only partially embed the MOF particles in order to leave part of them available to the gases in the environment, hence facilitating gas diffusion through them into the portion submerged into the CA, whilst assuring MOF immobilization on the substrate. In this way, a compromise between adhesion of the MOF to the substrate and accessibility of gas molecules was achieved. Total immersion of MOF particles into CA would probably further increase film stability, but with the undesired outcome of hindered access of gas molecules towards the interior of the sensing material.
To analyze whether the pressures applied during the soft-imprinting process affected the crystallinity of the MOF, we compared the PXRD patterns of MOF activated (desolvated) at 493 K under vacuum for 5 h with its corresponding films at 2, 4, and 6 bar. In Figure 6
, extreme cases of low (2 bar) and high (6 bar) pressure are shown. As can be seen, in both cases the PXRD patterns were very similar, despite the peaks shift verified in the diffractograms of the two films relative to the MOF powder, which corresponds to an experimental misalignment due to the distinct morphology of the analyzed samples (films and powder). These observations indicating that the crystallinity was retained after the soft-imprinting process, and no observable differences occur in the preparations between 2 and 6 bar. With this outcome, the suitability of soft-imprinting as a technique for the processing of MOFs onto solid substrates is confirmed.
2.3. Explosive Vapor Sensing
MOF powder was first deposited on glass substrates using either double-sided tape or the adhesive residue left after peeling off the tape. Similar procedures for the creation of MOF films can be found in the literature [12
]. We discarded double-sided tape as method to adhere the MOF to the substrate due to the strong photoluminescence in the region of 400–500 nm, coinciding with the emission band of the MOF (Figure S2
). Besides, solvents and other compounds contained in the adhesive tape may interact with the sensing material, altering its response towards the analytes to be detected. These drawbacks were minimized by using only the adhesive residue left by the tape, which showed no photoluminescence (Figure S2
) and allowed the fabrication of uniform MOF thin films. These films were exposed to DNT, resulting in an important quenching of their fluorescence (Figure S3
). This proves the potential of the developed samples as an explosives sensor, in agreement with the results published by other authors [12
]. However, the adhesive-residue method for the deposition of the MOF showed very little stability, as MOF particles were only partially adhered to the substrate. Even with extreme care, manipulation of the films led to MOF particles dropping onto every surface. Successive introduction and removal of the films from the fluorimeter caused a progressive loss in fluorescence intensity due to a decrease in the amount of surface covered by MOF particles that was even visible to the naked eye (Figure S4
). This instability limits the possibilities of the films to be used as sensors in commercial applications.
As an alternative, soft-imprinted MOF films on CA-coated quartz were used for the detection of DNT. Films prepared at 4 bar were chosen for this purpose. Photoluminescence spectra of activated MOF powder deposited by soft-imprinting showed an intense emission band at 460 nm when excited at 280 nm (Figure 7
, red line). This excitation wavelength was chosen based on the excitation spectra (Figure 7
, blue line), which showed an excitation region from 250 nm to 370 nm. The excitation at 280 nm provided the highest emission among the different wavelengths tested, hence maximizing the intensity of the band to be monitored in subsequent sensing experiments. Identical results were obtained by depositing MOF powder on glass with the adhesive residue left by a double-sided-tape. The high emission intensity observed in the photoluminescence spectrum of the sample suggests its suitability as a fluorescent probe.
The exposure of the films to DNT vapors resulted in a substantial quenching of the typical fluorescence of [Zn2
(bpee)] (Figure 8
). This quenching has been attributed to the presence of two -NO2
electron-withdrawing functional groups in the structure of DNT. As a result, the nitroaromatic π system would be electron-depleted and the molecule would behave as a π acceptor upon interaction with the organic ligands of the MOF [20
]. We quantified the quenching of fluorescence (%) as
, in which
is the maximum fluorescence intensity of the sample and
is the corresponding maximum intensity after exposure to DNT (Figure 8
, inset). The emission intensity was monitored at the wavelength of maximum emission (460 nm). Mean fluorescence quenching for a 10-s exposure was found to be 15%, indicating the fast response of our sensor towards DNT. On average, all films responded equally to the nitroaromatic irrespective of the pressure applied during the soft-imprinting procedure (Figure S5
), which suggests an elevated reproducibility of the method. However, it must be noted that this quenching was possibly underestimated due to experimental constraints. DNT exhibits extremely low vapor pressures, requiring long times for the generation of saturated environments. In our case, saturation was ensured by generating DNT headspace in a small vial during a long period of time (four days). However, the dilution of the saturated headspace during the manipulation of the vial for the introduction of the film cannot be discarded. This would explain different quenching values found by other authors [12
], besides other experimental variations or the existence of undisclosed preparation details in other publications that might affect the sensing capabilities of the MOF or its interaction with DNT.
Our DNT sensor benefited from the good sensing capabilities shown by the MOF and from the enhanced stability of the films prepared by soft-imprinting. This outcome confirms the suitability of [Zn2
(bpee)] for the detection of nitroaromatics. Besides, the soft-imprinting technique emerges as an ideal tool for the controlled and reproducible processing of MOFs or similar materials onto solid substrates. In our view, these promising results pave the way for an improved method for the fabrication of MOF sensors. To do this, the soft-imprinting technique still needs to be refined by analyzing all variables in the process and by extending the scope of this study to other MOFs and their exposures to other analytes. Besides, efforts are being made to establish a recovery procedure for the reutilization of the sensors. Meanwhile, the potential applications of our films are oriented towards disposable sensors. Last but not least, the system for the exposure to nitroaromatics needs to be perfected. Due to their low volatility, the typical procedure found in the literature merely consists of the insertion of the sensor into a sealed vial containing a saturated headspace generated by the desired compound [12
]. With this technique, which we also followed in our experiments, the concentration of nitroaromatics cannot be controlled. Our current experimental efforts are working towards the implementation of a system for the generation of controlled atmospheres of nitroaromatics, intended to be able to quantify the responses of our sensors.