**1. Introduction**

Metal organic frameworks (MOFs) constitute a group of materials consisting of the repetitive assembling of organic linkers and cluster metal ions. The resulting crystalline structures have a large specific surface area and a well-defined porosity. Their physicochemical properties can be tailored by selecting the nature and combination of both components yielding numerous MOFs [1,2]. MOFs have gained growing importance in many applications, such as gas storage [3,4], sensing [5,6], membrane processes [7], drug delivery [8,9], and environmental remediation [10–13]. In this last field, MOFs have been investigated in water purification, especially as: (i) adsorbents for the removal of hazardous pollutants [14–16], (ii) catalysts for the degradation of contaminants by Advanced Oxidation Processes (AOPs), such as heterogeneous Fenton and photo-assisted treatments [17–19], and (iii) membrane materials for the separation of toxic substances from water streams [20]. The growing interest in these applications is leading researchers to develop MOFs with high stability in aqueous medium.

The literature describes examples of unstable MOFs that lose their structure and pore network when exposed to water, such as MOF-5, while others showed stability in water, even after months, or have not been fully tested in this respect [21,22]. The stability of MOFs can be related to an assembly–disassembly equilibrium where both the electronic and steric e ffects between the linker and the metal cluster play an important role. Several studies have shown that the instability under wet conditions depends on the metal–linker bond strength [23,24]. Using highly basic ligands or acid metals results in much stronger bonds and greater stability in water, as in the cases of ZIF-8 or Al-MIL-53 [25]. In addition, the linker structure, the nature of the metal and the coordination of the clusters also play an important role. Thus, hydrophobic linkers, inert metal cations and

high-coordination clusters usually improve water stability. Moreover, high-charge metal ions lead to a stronger oxo-philic tendency that significantly strengthens the metal–linker bonds and improves the chemical stability of MOF [21]. Group-IV-metal- (Ti, Zr, Hf) based MOFs are examples of high-charge metal oxo-clusters, yielding abundant MOF structures with di fferent coordination numbers and structural diversity [26–29]. Ti-based MOFs have received grea<sup>t</sup> attention because of their redox activity (Ti<sup>3</sup>+/Ti4<sup>+</sup> transition capacity), photochemical property and biocompatibility [30]. Those based on Ti-carboxylate complexes have high stability and their functional properties can be improved by organic functionalization, mainly by amine groups [31,32]. The degradation of MOFs exposed to water can occur by two di fferent mechanisms, ligand displacement and hydrolysis [25]. The ligand displacement involves the exchange of a ligand by a water molecule, leading to the hydration of the metal and ligand lixiviation; while, upon hydrolysis, water dissociation occurs at the metal centers, the dissociated OH groups coordinate to the metal clusters and the metal–ligand bond is broken [33].

Nevertheless, predicting the water stability of MOFs remains qualitative and the literature is controversial, mainly due to the methodology used to assess the stability. The simplest process consists of exposing the MOF to water and comparing its properties prior and after a given contact time, mainly by weight loss, X-ray di ffraction and N2 adsorption–desorption at −196 ◦C [29]. Water adsorption–desorption isotherms are also useful since they provide information about the hydrophilic–hydrophobic character of MOFs [34]. Additionally, however, if the MOF is being considered for application in the aqueous phase, it should also be investigated whether the sample is partially dissolved [29,35]. Recently, Taheri et al. [36] performed a quantitative study of the ligand and Zn2<sup>+</sup> released when ZIF-8 was immersed in water. They observed that the lixiviation of both species occurred during the first hour and after 24 h an equilibrium was achieved. No other studies deal with the analysis of the filtrate instead of the remaining solid in the literature. Lixiviation from the MOF must be considered for the sake of application in water remediation since the released metals and/or ligands can represent some hazards for the environment.

The literature contains examples concerning the post-modifications of MOFs to improve their water stability. Wen et al. [37] coated NH2-MIL-125(Ti) with a siloxane that increased its dye adsorption capacity while maintaining its crystal structure and morphology. Composites based on UiO-66-NH2 and a nitrile butadiene rubber sponge are more stable in water than the bare MOF because the sponge provides a hydrophobic character that prevents hydrolysis. These composites were shown to be efficient adsorbents for 2,4-dichlorophenoxyacetic acid removal from water and were easily recoverable thanks to its sponge conformation [38]. An interesting study has recently been reported, describing the improved water stability of ZIF-8 by growing it along with UiO-66-NH2, giving rise to a core–shell UiO-66-NH2@ZIF-8 hybrid. This heterostructure showed excellent water stability, high adsorption capacity and selectivity for the separation of rare earths from water compared to other adsorbents [39]. Non-amine UiO-66 has been also incorporated in a composite membrane based on graphene oxide (GO) with a high water stability and highly e fficient for dyes and antibiotics separation [40]. Although interesting, many of these modifications are not so easy to implement in technical applications.

Here, we focus the attention on NH2-MIL-125(Ti), a Ti-oxo-cluster MOF, due to its notable interest in water treatment by photocatalysis, a well-known AOP for the removal of di fferent pollutants [1,41–43]. In a previous study, we demonstrated its high photocatalytic e fficiency for the removal of some emerging pollutants and its stability during the reaction, comparing its structural, textural and morphologic properties before and after use [44]. However, in a preliminary test, additional organic matter was detected in the aqueous medium after reaction, thus suggesting that certain partial dissolution of the solid occurred. Thus, in this work, we initially studied the water stability of NH2-MIL-125(Ti) through a systematic analysis of the aqueous phase. Then, in order to improve its stability, NH2-MIL-125(Ti) was subjected to various thermal treatments. In particular, the study highlights that the stability of NH2-MIL-125(Ti) in water can be successfully improved, providing useful information for water-related applications.

#### **2. Results and Discussion**

#### *2.1. Stability of the NH2-MIL-125(Ti)*

NH2-MIL-125(Ti) is an amine-functionalized isostructure of the MIL-125(Ti) formed by both octahedral and tetrahedral cages [45,46]. It is composed by titanium oxo-clusters and amino-terephtalate, both providing a high density of hydrophilic sites where water molecules can be adsorbed [47]. We had previously synthesized NH2-MIL-125(Ti), doped and successfully used as solar-light driven photocatalyst [44]. Although its structural stability under the reaction conditions was confirmed by X-ray di ffraction, scanning electron microscopy and N2 adsorption–desorption at −196 ◦C, some ligand release was detected in the aqueous medium. To the best of our knowledge, there are no reports about the amount of linker leached when NH2-MIL-125(Ti) is put in contact with water. Thus, to determine the amount of released linker, NH2-MIL-125(Ti) was suspended in aqueous medium and the filtrates were analyzed, the leachate percentage being determined by:

$$\text{Leachate}\left(\%\right) = 100 \times \frac{\text{C}\_{\text{linker}}}{\text{C}\_{\text{NH}\_2\text{-MLL}\cdot125(\text{Ti})}} \cdot \frac{\text{M}\_{\text{NH}\_2\text{-MLL}\cdot125(\text{Ti})}}{6 \times \text{M}\_{\text{linker}}} \tag{1}$$

where Clinker is the linker concentration dissolved in the liquid phase (mg·L−1), CNH2-MIL-125(Ti) is the concentration of the MOF suspended in water, MNH2-MIL-125(Ti) and Mlinker are the molecular weight values of the MOF (1653.74 <sup>g</sup>·mol−1) and the linker (2-amino benzene dicarboxylate, NH2-BDC, 179.12 <sup>g</sup>·mol−1), respectively. Figure 1a shows the evolution of dissolved linker and the corresponding leachate percentage upon contact time. NH2-MIL-125(Ti) undergoes relatively high linker leaching in water, which increases significantly over time to reach 40 mg·L−<sup>1</sup> after 24 h, about 25% of the initial linker content of bare NH2-MIL-125(Ti). This leaching process occurs continuously over time, without reaching equilibrium. Similar behavior has recently been reported for ZIF-8, whose lixiviation started during the first hour and required one day to achieve equilibrium [36]. This leaching is detrimental regarding the potential water-related applications of NH2-MIL-125(Ti). For the sake of improving the stability of NH2-MIL-125(Ti) by reducing ligand leaching as much as possible, this MOF was subjected to di fferent thermal treatments at di fferent temperatures and holding times and under di fferent atmospheres. The temperature was selected after studying the thermal behavior of this MOF by thermogravimetric analysis in air (Figure 1b). NH2-MIL-125(Ti) is thermally robust and su ffers a strong weight loss of 54% within the 300 to 350 ◦C range, due to the oxidation of the organic linkers [36,48]. After increasing the temperature up to 500 ◦C, a small weight loss (9%) occurs, corresponding to the removal of the hydroxo-groups in the metal oxo-clusters, giving rise to TiO2, in a similar way to the analogous non-amine MIL-125(Ti) [46,49]. Based on this analysis, the 150–300 ◦C range was selected for the thermal treatment addressed to stabilize the NH2-MIL-125(Ti) without disturbing its structure and composition.

**Figure 1.** (**a**) Time course of linker concentration and leachate percentage in water from NH2- MIL-125(Ti) upon contact time; (**b**) thermogravimetric analysis in air of NH2-MIL-125(Ti) and its corresponding derivative.

#### *2.2. Vacuum Treatment*

During the synthesis of MOFs, excess ligand and solvent molecules can remain trapped in the pores of the framework, which can be detrimental for their future applications. In some cases, vacuum drying may be sufficient to purify the MOF, although it can lead to lower surface areas than expected due to the partial collapse of the structure [35]. NH2-MIL-125(Ti) was thus subjected to vacuum treatment in a temperature range of 100 to 300 ◦C for 16 h to remove the excess linker and solvent molecules. Figure 2 shows the X-ray diffraction (XRD) patterns of the modified solids compared with that of the bare NH2-MIL-125(Ti). In the notation, "V" refers to vacuum treatment, and the numbers represent the temperature (◦C) and time in hours of the vacuum treatment. As can be seen, this treatment does not modify the crystalline structure of the MOF until reaching 300 ◦C. At this temperature, the NH2-MIL-125(Ti) structure collapses, and the resulting material does not describe any crystalline XRD profile, appearing as an amorphous material. Before that, a reduction in the peaks' intensity as the temperature increases is evidenced, indicating a gradual loss of crystallinity. Thus, the degree of crystallinity and the crystal size, as determined by the methodologies described in Section 3.2, are collected in Table 1. The greatest reduction on crystallinity was observed when reaching 250 ◦C. However, since NH2-MIL-125(Ti) is an amine-functionalized isostructure of the MIL-125(Ti), both materials would show the same XRD profile, so it would be necessary to corroborate that the amine group in the ligand is maintained. The NH2-BDC presence was confirmed by UV-vis spectroscopy (F1 in the supplementary information). Vacuum-treated samples showed two absorption bands, at 280 and 370 nm, due to the absorption of Ti-oxo-clusters and NH2-BDC linker, respectively. Both bands characterize the NH2-MIL-125(Ti) and differ from the spectrum of MIL-125(Ti), which only has one band due to the absorption on Ti-oxo-clusters [49,50]. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images of V-250-16 were taken and compared with those from the original sample (Figure 3). The MOF particles show a thin and disk-like shape with an average size close to 500 nm, very similar to those of the NH2-MIL-125(Ti), which confirms that the vacuum treatment does not significantly modify the morphology of the MOF.

**Figure 2.** X-ray diffraction (XRD) patterns of the NH2-MIL-125(Ti) treated under vacuum at different temperatures for 16 h. The original NH2-MIL-125(Ti) pattern is included as reference.

**Table 1.** Crystallinity percentages and crystal size values of NH2-MIL-125(Ti) before and after vacuum treatment at different temperatures for 16 h.


Relative crystallinity to that of the original MOF.

1

**Figure 3.** Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images of the original NH2-MIL-125(Ti) and V-250-16 sample.

Figure 4 shows the −196 ◦C N2 adsorption–desorption isotherms of the NH2-MIL-125(Ti) subjected to the vacuum treatments and Table 2 summarizes the porous textural characteristics derived from those isotherms. Most of the solids treated under vacuum showed isotherms similar to that of the original NH2-MIL-125(Ti), indicative of predominantly microporous texture with some relative contribution of mesoporosity and fairly high surface area values [39,44,48]. Only the V-300-16 shows remarkable differences, with very low N2 adsorption and SBET due to the linker oxidation and structure collapse. It is worth noting that an increase in the temperature, below 300 ◦C, is associated with some slight increase in the microporous surface, which can be related to the removal of adsorbed species, most probably excess linker and/or solvent molecules, not removed during the washing process that block the pore network [35]. According to these results, 250 ◦C appears to be the best temperature to carry out the vacuum-stabilization of the NH2-MIL-125(Ti), maintaining its structure and porosity.

**Figure 4.** N2 adsorption–desorption isotherms at −196 ◦C of the original NH2-MIL-125(Ti) and after treatment under vacuum at different temperatures for 16 h.

**Table 2.** Porous texture characterization of the NH2-MIL-125(Ti) and after treatment under vacuum at different temperatures for 16 h.


1 SBET, specific surface area; 2 SMP and 3 SEXT, microporous and non-microporous surface area; 4 VT and 5 VMP, total and micropore volume, respectively.

Figure 5 depicts the evolution of the linker concentration analyzed in the liquid phase and the corresponding leaching percentage, upon contact time in water. The vacuum thermal treatment allows for significantly reducing the amount of linker leaching, that effect being more pronounced at a 200–250 ◦C treatment temperature. However, despite this reduction, there is still a considerable amount of linker in the liquid phase, corresponding to a 15% leaching, and it is also noteworthy that the lixiviation does not reach an equilibrium, even after 24 h. Therefore V-200-16 or V-250-16 samples cannot still be considered as stable materials regarding potential applications in the aqueous phase.

**Figure 5.** Linker leaching in water from NH2-MIL-125(Ti) before and after vacuum heat treated upon contact time.

#### *2.3. Thermal Treatment in Air Atmosphere*

Figure 6 depicts the XRD diffractograms of the NH2-MIL-125(Ti) subjected to calcination during 16 h at different temperatures in air-circulating muffle furnace. As previously observed in the vacuum treatment, the NH2-MIL-125(Ti) structure was maintained up to 250 ◦C, while it collapses at 300 ◦C, disappearing all the characteristic peaks of NH2-MIL-125(Ti). The degree of crystallinity and the crystal size of each sample are collected in Table 3. Increasing the temperature decreases the crystallinity of the samples, accompanied by a reduction in crystal size. These modifications can be related to the loss of solvent and excess linker molecules that are usually trapped in the MOF cages [35]. To learn more about the effect caused by air heating, additional treatments were conducted at 250 ◦C, but extending the treatment time. Figure 7 compares the diffraction patterns of the samples heated for 16, 48 and 72 h, where it can be seen that the characteristic peaks of the original structure are still remaining, but showing decreasing intensities as the calcination time increased, thus indicating a considerable gradual loss of crystallinity, reaching about 60% after 72 h, which means a certain amorphization of the material.

**Table 3.** Crystallinity percentages and crystal size values of NH2-MIL-125(Ti) before and after heat treatment in air under different conditions. The values for M-250-48 after 24 h in contact with water are included.


Relative crystallinity to that of the original MOF.

1

**Figure 6.** XRD diffraction patterns of NH2-MIL-125(Ti) heated in air at different temperatures for 16 h. The original NH2-MIL-125(Ti) pattern is included as reference.

**Figure 7.** XRD diffraction patterns of NH2-MIL-125(Ti) heated in air at 250 ◦C for 16, 48 and 72 h. The original NH2-MIL-125(Ti) pattern is included as reference.

**θ** Once again, the presence of NH2-BDC in all heat-treated samples was confirmed by UV-vis spectroscopy (Figure S2 in the supplementary information). The UV-vis spectra showed the two absorption bands characteristic of the NH2-MIL-125(Ti), unlike the single band that characterizes the MIL-125(Ti) [49,51,52]. Figure 8 collects SEM and TEM images of some calcined samples that can be compared with those of the original NH2-MIL-125(Ti) shown in Figure 3. As can be clearly observed, the morphology of M-300-16 is very different. The well-defined disk particles of the NH2-MIL-125(Ti) disappeared and became an amorphous morphology, corroborating the structural collapse described

[ ]

above. However, the morphology and disk-shape of the original MOF remained after thermal treatment at 250 ◦C for 48 h, as well as the particle size (≈ 500 nm).

**Figure 8.** SEM image of M-300-16. SEM and TEM images of M-250-48.

Figure 9a depicts the −196 ◦C N2 adsorption–desorption isotherms of NH2-MIL-125(Ti) subjected to thermal treatments in air and the derived porous textural characteristics are summarized in Table 4. As in the case of heat treatment under vacuum, these samples showed an isotherm typical of microporous solids [39,44,48]. Again, the calcination in air at 300 ◦C caused the loss of structure and the collapse of the pore network, resulting in a N2 adsorption–desorption isotherm of a non-porous material. Again, in the range below 300 ◦C, increasing the calcination temperature produces some slight increase in the microporous surface. However, in the sample treated at 250 ◦C, the observed increase in surface area corresponds to the non-microporous one. This effect can be related with the decrease in crystallinity and increased disorder observed by XRD. When comparing the vacuum and the air thermal treatments, no obvious differences were detected, so it seems evident that temperature is the main factor governing the purification process. Regarding the effect of the treatment time, the N2 adsorption–desorption isotherms of their corresponding samples remain typical of microporous solids (Figure 9b). Increasing the treatment time led to some moderate reduction in the BET surface area, affecting to the microporous range, while the non-microporous area was increased, probably due to the loss of crystallinity seen by XRD. Therefore, thermal treatment at moderate temperature (≈250 ◦C) allows purifying and improving the stability of the tested MOF, while maintaining its porous structure and surface area close to 1000 <sup>m</sup>2·g<sup>−</sup>1, analogous to other untreated NH2-MIL-125(Ti) [41–44].

**Figure 9.** N2 adsorption–desorption isotherms at −196 ◦C of the original NH2-MIL-125(Ti) and after heat treatment in air at different temperatures for 16 h (**a**) and at 250 ◦C for different times (**b**).


**Table 4.** Porous texture characterization of NH2-MIL-125(Ti) after heat treatment in air under different conditions.

1 SBET, specific surface area; 2 SMP and 3 SEXT, microporous and non-microporous surface area; 4 VT and 5 VMP, total and micropore volume, respectively.

Figure 10 shows the results of the stability in water of the samples subjected to calcination in air at different temperatures and times. Thermal treatment in air for 16 h stabilizes the NH2-MIL-125(Ti) and the leachate percentage diminishes at increasing temperature. At 250 ◦C, this improvement of stability in water is significantly higher than the observed with the thermal vacuum treatment (see Figure 5) and now a quasi-equilibrium state is achieved. Extending the air thermal treatment up to 48 h reduces the leaching percentage to less than 7% and maintains the quasi-equilibrium state, although no further improvement was observed at higher treatment times. Characterization of the M-250-48 sample after water exposition for 24 h was carried out to check possible structural and textural changes (Figure 11). There were no significant differences in X-ray diffractograms or N2 adsorption–desorption isotherms before and after contact with water during the above-mentioned time. The SBET slightly decreased somewhat, from 1046 to 970 <sup>m</sup>2·g<sup>−</sup>1, but the crystallinity remained unchanged (see values in Table 3; Table 4). The NH2-BDC also remained in the structure, including the presence of the amine group, as confirmed by its UV-vis spectrum (Figure S3 in the supplementary information) that presents the characteristic absorption bands of this linker. These results demonstrate that it is possible to purify and stabilize the NH2-MIL-125(Ti) MOF by a fairly simple thermal treatment in air at 250 ◦C for 48 h, without affecting its structure and porous texture.

**Figure 10.** Linker leaching in water from the original NH2-MIL-125(Ti) and after heat treatment in air for 16 h at different temperatures and at 250 ◦C during different times.

**Figure 11.** (**a**) XRD patterns and (**b**) N2 adsorption–desorption isotherms at −196 ◦C of M-250-48 before and after contact with water for 24 h.

#### **3. Materials and Methods**

### *3.1. NH2-MIL-125(Ti) Synthesis*

NH2-MIL-125(Ti) was prepared following the methodology reported in our previous work (chemical information included in Figure S4 in the supplementary information) [44]. Typically, 2-amino benzene dicarboxylic acid (6 mmol, NH2-BDC, Sigma Aldrich, Madrid, Spain, 99%) was dissolved in dimethylformamide (25 mL, DMF, Sigma Aldrich, Madrid, Spain, ≥ 99.8%) under stirring. Titanium isopropoxide (3 mmol, Sigma Aldrich, Madrid, Spain, ≥ 98%) was dropped onto the mixture and stirred until homogenized, after which methanol (25 mL, CH3OH, Sigma Aldrich, Madrid, Spain, anhydrous 99.8%) was added. The whole mixture was kept in agitation for 30 min, drawn off to a 65 mL Teflon autoclave and placed in an oven at 150 ◦C for 16 h. The resulting solid was recovered by centrifugation (5 min, 5000 rpm) and washed three times with DMF (100 mL, 30 min) and three times with methanol (100 mL, 30 min), recovering the solid in all cases by centrifugation. The final drying of the solid was performed overnight at 60 ◦C.

NH2-MIL-125(Ti) was subjected to different thermal treatments. The first series was obtained by heating the NH2-MIL-125(Ti) at 150, 200, 250 and 300 ◦C for 16 h under vacuum using a degassing station (Micromeritics VacPrep 061, Norcross, GA, USA) attached to the TriStar 123 equipment. For the second series, the NH2-MIL-125(Ti) was heated in a muffle furnace (Fuji Electric, MOD 12 PR/400, Barcelona, Spain) in air at 150, 200, 250 and 300 ◦C for 16 h, and also at 250 ◦C for 48 and 72 h. The resulting samples were denoted as X-T-t, where X indicated the type of treatment, V for vacuum and M for muffle furnace, T was the treatment temperature and t the treatment time.
