Shuai-Liang
Yang
a,
Gen
Li
a,
Xiao-Yan
Liu
a,
Lin
Zhang
a,
Yan
Xu
a and
En-Qing
Gao
*ab
aShanghai Key Laboratory of Green Chemistry and Chemical Processes, School of Chemistry and Molecular Engineering, East China Normal University, Shanghai 200062, China. E-mail: eqgao@chem.ecnu.edu.cn
bInstitute of Eco-Chongming, Shanghai 202162, China
First published on 5th July 2022
The stability of metal–organic frameworks (MOFs) in solutions is of essential importance from the perspectives of scientific research and practical applications. Powder X-ray diffraction (PXRD) with isolated solids is a common protocol to judge the stability. However, solid-state characterization may lead to ignorance of possible chemical leaching and even partial collapse, and the ignorance may lead to confusing or even false conclusions about stability and relevant properties. Herein, based on the pH-responsive fluorescence associated with alkaline hydrolysis of MOFs, a fluorometric method is proposed for characterizing the stability of luminescent or luminogenic MOFs in basic solutions. Exemplified by two Zr(IV) MOFs, Zr–TCPBP (tetracarboxylate linker) and UiO-66-NH2 (dicarboxylate), the method has been validated by a comparative study with solid-state characterization using PXRD, scanning electron microscopy (SEM) and thermogravimetry. The method has the advantages of facile operation, high sensitivity and good complementarity to PXRD and SEM. It can be used to continuously monitor the whole hydrolytic process starting with nondestructive local hydrolysis and ending with abrupt and destructive hydrolysis, giving the pH values for different hydrolytic stages. The method has great potential to be extended for characterizing the stability of MOFs towards acids or other substrates in solution.
Luminescent or luminogenic MOFs dispersed in solution often show sensitive emission changes in response to some chemical species. This has encouraged extensive studies to develop MOF-based chemical sensors with excellent sensitivity and selectivity having been reported.30–35 Most studies attributed the fluorescence response to intact MOFs, the intactness being judged from PXRD with solids isolated from sensing media. The issue that the luminescence of MOF dispersions can be strongly affected by a minute amount of exuded fluorescent species is rarely addressed. The overlook may lead to erroneous conclusions and understandings.
Zr(IV) MOFs built from Zr6O4(OH)4 nodes and carboxylate-terminated linkers are deemed to be a class of highly stable MOFs.36,37 High acid resistance at pH < 0 has been commonly reported,38–41 which is understandable considering the well-matching double bonding of carboxylate groups with high oxophilic Zr6 clusters and the low pKa of aromatic –COOH.27 However, the stability in basic solutions has been reported with great discrepancy even for the same MOF. Taking UiO-66-NH2 as an example, Lillerud et al. reported that the MOF is completely decomposed into an amorphous phase after treatment with a pH 14 solution.42 According to Van Der Voort et al., the MOF maintains a good crystalline structure after placing in a pH 12 NaOH solution for two months.43 However, Aguilera-Sigalat et al. showed that the MOF undergoes certain structural damage in alkaline solution with pH > 9.44 These quite different conclusions are all based on PXRD analysis, reflecting the inadequacy of PXRD in characterizing the stability of MOFs in liquid media.
Herein, we developed a fluorometric method for determining the stability of MOFs in basic solutions. The method is based on the fluorescence response induced by alkaline hydrolysis of MOFs. By a comparative study combining PXRD, scanning electron microscopy (SEM) and thermogravimetry (TG), the feasibility and reliability of the method are first demonstrated with a Zr(IV) MOF (Zr–TCPBP) with 2,2′,6,6′-tetra(4-carboxyphenyl)-4,4′-bipyridine (TCPBP)45 and then extended to UiO-66-NH2 (Scheme 1). The method has the advantages of simple operation, high sensitivity and high informativeness, representing a nice complement to conventional solid-state characterization methods. In particular, it can be used to monitor the whole process from minor leaching of linkers to obvious hydrolysis and to complete hydrolysis, giving the pH range of rapid decomposition.
On the one hand, the intensity initially increases smoothly, then abruptly rises in the range of pH 11.0–11.8, and finally becomes constant beyond the range (Fig. 1b). The critical point for the abrupt transition is at about pH 11.4, as estimated from the derivative curve. The intensity transition means that a small pH change can lead to a significant fluorescence enhancement, which indicates the potential application of the MOF as an ultrasensitive pH probe working in the highly basic range. On the other hand, the λmaxvs. pH plot shows two linearly dependent regions, a slow blue shift for pH varying from 7.5 to 9.6 and a rapid blue shift for pH 9.6–11.5 (Fig. 1c). The linear variation indicates the feasibility of quantitative pH determination in the basic range by monitoring λmax. Fluorescence pH sensing based on wavelength shifts, compared to that based on intensity changes, can have some advantages such as high anti-interference and visual color changes.46 Converting the emission spectra to chromaticity coordinates in the Commission Internationale de L'Eclairage (CIE) chromaticity diagram gives a linear region for pH 7.5–11.5, which corresponds to the cyan-to-blue change in the fluorescence color (Fig. S1†).
To explore the mechanism of the complex pH dependence of fluorescence in basic solutions, we filtered the Zr–TCPBP dispersions in pH 7.2, 9.6 and 11.5 solutions and measured the fluorescence of the filtrates and solid residues re-dispersed in water. The filtrate for the nearly neutral dispersion shows no emission, but those for pH 9.6 and 11.5 show significant emission, which is even more intense than the emission of the unfiltered dispersions (Fig. 2a). The latter two filtrates share the same λmax (395 nm) with the free organic ligand dissolved in basic solution, indicating the presence of the ligand in the filtrates and also in the corresponding dispersions. That is, there is no leaching of the ligand for neutral dispersions, but significant leaching occurs for the dispersions of pH 9.6 and 11.5 because of hydrolysis of the MOF. The emission intensity of the pH 11.5 filtrate was significantly higher than that for pH 9.6, confirming the expectation that the degree of leaching or hydrolysis increases with pH. As can be seen, the pH 11.5 dispersion and its filtrate both have λmax = 395 nm, and the emission intensity of the filtrate is very close to that of the ligand solution with a controlled concentration. This suggests almost complete MOF hydrolysis and ligand release at that pH. The fact that the emission of the dispersion is weaker than that of its filtrate can be simply because of the scattering effect of the solid particles in the dispersion. The conclusions are corroborated by analyzing the emission of the filter residues re-dispersed in neutral water. The residue from the pH 9.6 suspension shows almost identical emission to the as-synthesized MOF (Fig. 2b), suggesting that the framework is preserved with minor ligand leaching. However, the residue from the pH 11.5 dispersion is essentially non-fluorescent and shows no obvious IR bands attributable to the ligand (Fig. S2†), indicating almost complete hydrolysis of the MOF to zirconium hydroxide hydrate at the high pH.27
Based on the above analysis, the fluorescence variation of the MOF dispersion with pH can be well justified as follows. The pH-independent fluorescence in the pH 4.4–7.2 range comes from the MOF itself without ligand leaching. The fluorescence observed at an intermediate pH (7.5 < pH < 11.5) is an envelope of the emissions from the leached organic ligand and the retained MOF. The blue shift and intensity enhancement with pH are due to the short-wavelength and high-intensity contribution from the ligand becoming more and more significant. At pH > 11.8, the constant fluorescence is related to complete release of the ligands as a result of complete hydrolysis.
The pH-induced fluorescence change associated with partial to full basic hydrolysis inspired us to propose a fluorometric approach for characterizing the stability of MOFs in basic solutions. The hydrolysis occurs simply owing to the competition between the carboxylate group and increasing OH− in binding metal ions at the node of the framework.27 For Zr–TCPBP, the small intensity increment below pH 11 indicates that a small portion of the ligands is replaced by OH−. The partial hydrolysis at nodes does not necessarily cause severe framework collapse, especially for the high-connectivity robust frameworks like Zr–TCPBP. The abrupt intensity rise above pH 11 indicates substantial ligand leaching because of the onset of the drastic destructive hydrolysis of the framework. The onset pH of the fluorescence change (ligand leaching) and the onset and critical pH for an abrupt fluorescence transition can be used as measures of the stability of the MOFs at different stages.
To verify the reliability of the fluorometric method, we examined the alkali stability of Zr–TCPBP by means of conventional solid-state characterization methods. The most widely used method is PXRD analysis with the solids isolated from dispersions. The method is unique for demonstrating the retention of bulk crystallinity. Zr–TCPBP shows no significant change in PXRD after being soaked at pH up to 11 (30 mg of MOF in 10 mL of aqueous solution) (Fig. 3a, pH 11*). This seems to indicate that the MOF is stable at pH 11, but according to fluorescence analysis, there is actually significant leaching of the ligand at that pH. The conclusion drawn from PXRD actually depends on the amount of the MOF relative to the solution used. For 30 mg of MOF treated with 100 mL of solution (the concentration of the MOF is the same as that for fluorescence tests), the PXRD for pH 10.5 treatment becomes weaker especially for high-angle peaks (Fig. 3a). For pH 11, the PXRD worsens and shows only two broadened low-angle peaks, which indicates the preservation of only some degree of long-range order in the MOF. The sample treated at pH 11.5 shows a profile typical of amorphous solids, suggesting complete hydrolysis of the MOF. The results are consistent with fluorescence analysis, both confirming partial to complete hydrolysis as the pH is increased. However, the PXRD analysis is less sensitive and cannot definitely disclose the ligand leaching below pH 9.5, whereas the fluorescence analyses disclose the onset of leaching (hydrolysis) at pH 7.5. Furthermore, the facile pH-fluorescence titration analysis reveals an abrupt hydrolytic transition starting at pH 11 and completing at pH 11.8, with a critical pH of 11.4. The abrupt transition is difficult to identify and quantify with the PXRD method. In brief, the fluorometric method can be a simple but useful complement to PXRD for characterizing the stability of MOFs in basic solutions and provides more precise and detailed information of ligand leaching and framework decomposition.
SEM was used to analyze the morphology of the samples treated at different pH values (the concentration of the MOF is the same as that used in the fluorescence test). The as-synthesized sample of Zr–TCPBP contains uniform cubic crystallites of size about 400 nm (Fig. 4). There are no significant changes in the size and morphology after pH 9.5 treatment. This, in accordance with PXRD, confirms that the minor ligand leaching indicated by fluorescence analysis hardly affects the exterior appearance and the interior crystallinity, giving a framework with linker-vacancy defects. In other words, SEM is also less sensitive than the fluorometric method, which can detect a small degree of nondestructive local hydrolysis at Zr6 nodes without framework collapse. At pH 10.5, the SEM images show a mixture of cubic, irregularly shaped and rough-surfaced particles of different sizes, indicating significant hydrolysis of the MOFs. At pH 11.5, the image is predominated by small, inhomogeneous and shapeless particles, confirming almost complete hydrolysis of the MOF. The morphological changes at pH 10.5 and 11.5 are consistent with the results obtained from PXRD and fluorescence analyses.
Thermal gravimetry (TG) was performed with the isolated solids to confirm the ligand leaching (Fig. S3†). The TG plots of Zr(IV) MOFs in air flow generally feature a plateau ascribable to the “dry” frameworks with dehydrated Zr6O6 nodes and multicarboxylate linkers, followed by rapid weight loss owing to linker decomposition and finally by a constant residual weight corresponding to ZrO2.42,47 The plateau/residual weight ratio can be used to estimate the linker/metal molar ratio. For pristine Zr–TCPBP, the TCPBP/Zr ratio was estimated to be 0.51, very close to the 1:2 ratio expected for a defect-free structure. The linker/metal ratios for treated solids were plotted against pH at which the MOF was treated (Fig. 3b). As can be seen, the ratio does not change after treatment at near-neutral pH. The treatment at pH 9.5 (10.6) leads to a slight (obvious) decrease of the ratio, and an abrupt drop was witnessed for increasing pH from 10.6 to 11.7. The trend confirms the ligand leaching process disclosed by fluorescence analysis.
As mentioned earlier, the literature studies are divided based on the basic pH range in which UiO-66-NH2 is stable.42–44 Based on the above investigation with Zr–TCPBP, we believe that the discrepancy mainly arises from the different conditions for treating the MOF, especially the relative amount of the MOF to the solution, which has attracted little attention. We reassessed the basic hydrolytic stability of UiO-66-NH2 by a combined study with fluorescence analysis, PXRD and SEM.
The emission spectra of UiO-66-NH2 dispersed in aqueous solutions with pH 6.9–10.8 (0.1 mg mL−1) are shown in Fig. 5a. The emission witnesses no shift in λmax (425 nm) but a significant enhancement in the intensity with increasing pH. The intensity versus pH plot shows a slow increase for pH varying from 6.9 to 9.2, followed by an abrupt rise for pH 9.2–10.5, with the inflection point at pH 10.3 (Fig. 5b). The trend is similar to that observed for Zr–TCPBP, but the rapid intensity transition occurs at lower pH. This could indicate a low alkali resistance for UiO-66-NH2. It is worth mentioning that a similar rapid fluorescence enhancement induced by a pH increase has been reported previously for UiO-66-NH2 and also for its post-synthetically modified derivatives.44 However, the authors ascribed the enhancement to the formation of the –NH− group by deprotonating the amino groups within the MOFs. To determine the real emissive species and the enhancement mechanism, the spectra of the supernatants obtained by filtering the dispersions were recorded (Fig. S4†). The spectra show the same λmax as those observed for the dispersions, and the variation of emission intensity with pH is very similar to that for the dispersions (Fig. 5b). Notably, even at pH 6.9, the supernatant is fluorescent, with the intensity being slightly lower than that of the unfiltered suspension (Fig. S5†). The results strongly suggest that the emission predominantly arises from the linker leached to the liquid phase and that leaching occurs even at neutral pH. The emission from solid particles is minor. For further comparison, the aqueous solution of the linker, 2-aminoterephthalic acid, was investigated at various pH values. The results reveal that the linker at a fixed concentration shows pH-independent emission spectra in the range of pH 6.3–10.9, with λmax = 425 nm (Fig. S6†). The invariant emission suggests invariant emissive species in the wide pH range. Referring to the pKa values of comparable compounds (4.92 (–COOH) and 2.14 (–NH3+) for 2-aminobenzoic acid48 and 3.20 and 4.30 for terephthalic acid49), the existing form of 2-aminoterephthalic acid in the above pH range should be the divalent anion with two carboxylic protons removed. Beyond the range, the fluorescence of the free linker decreases at pH ≥ 11.5 and is almost completely quenched at 13.2. This phenomenon can be ascribed to the deprotonation transformation from –NH2 to –NH−, consistent with the long-known example that aniline becomes non-fluorescent at pH > 13 for the same reason.50 On the other hand, lowering the pH below 5.1 leads to a gradual red shift and intensity increase, which can be attributed to the gradual protonation of the divalent anion. Overall, the investigations with filtrates and the free linker definitely support that the alkali-induced gradual and then abrupt emission enhancement for UiO-66-NH2 is due to the increased linker leaching, or basic hydrolysis, rather than amino deprotonation. In particular, the fluorescence analysis reveals linker leaching even under neutral conditions and hydrolytic decomposition with a critical pH of 10.3. In addition, the kinetic measurements at various pH values suggest that the linker leaching and hydrolytic decomposition are fast and reach equilibrium within 10 min (Fig. S7†).
To validate and complement fluorescence analysis, PXRD and SEM were performed with the solid samples isolated after dispersing UiO-66-NH2 in neutral to basic solutions (the solid-to-liquid ratio is the same as for fluorescence analysis). As shown in Fig. 5c, the sample after dispersion at pH = 9 displays a well-defined diffraction profile, and the SEM image shows uniform and discrete crystallites (Fig. 5e). There are no significant differences in the crystallinity and morphology as compared with pristine UiO-66-NH2. The sample treated at pH 10 shows only a few weak diffraction peaks, and the particles become smaller and inhomogeneous and seem to be adhered together by the amorphous substance (Fig. 5f). For the sample treated at pH 11, SEM and PXRD confirm the formation of a featureless amorphous agglomerate (Fig. 5g). The above results confirm the rapid hydrolytic decomposition and ligand leaching above pH 9. The solid-state characterization methods give no information about ligand leaching at pH ≤ 9. Combining the dispersion fluorescence method and the solid-state methods, it can be concluded that partial ligand leaching (local hydrolysis at nodes) at a relatively low pH does not lead to significant degradation in the crystallinity and morphology. In addition, fluorescence titration analysis at high pH provides a facile method to determine the pH range for significant and complete hydrolytic decomposition.
Finally, it is worthwhile to emphasize the lessons learned from our study with Zr–TCPBP and UiO-66-NH2. First, we should be cautious in judging the stability of MOFs towards chemical species in solutions from PXRD with the isolated solids. The PXRD results are strongly dependent not only on the concentration of the chemical species but also on the solid-to-liquid ratio used for treating MOFs (see Fig. 3a for Zr–TCPBP and Fig. S8† for UiO-66-NH2). The latter factor has been largely neglected, which has led to inconsistent statements about the stability of MOFs. The PXRD method is incapable of detecting chemical leaching through nondestructive or even partially destructive processes. Second, fluorescence is sensitive to minor species. Many literature reports on fluorescence sensing have assigned the fluorescence (or stimulus-induced fluorescence change) of MOF suspensions to intact frameworks, based on stability characterization with PXRD.51–53 Unfortunately, as indicated by this work, solid-state characterization methods are far from sufficient to warrant the assignment. To avoid a false understanding of the underlying chemistry and physics, careful examination of the liquid phases is required for identifying the real emissive/responsive species.
Footnote |
† Electronic supplementary information (ESI) available: Experimental section and supplementary graphic materials, Fig. S1–S8. See DOI: https://doi.org/10.1039/d2qi00949h |
This journal is © the Partner Organisations 2022 |