Phase Transition of Metal–Organic Frameworks for Constructing Nanocomposite Materials

Through encapsulating functional materials, metal–organic framework (MOF) composites show extraordinary potential in various fields due to the excellent synergistic effects between the host and guests. However, many attracting functional species, such as enzymes, could be easily damaged during the synthesis of MOF composites. Herein we report a new strategy, namely pressure-amorphization-stimulation-recovery (PASR), in which crystalline MOFs were transferred to the amorphous MOFs at certain mechanical pressure, followed by recrystallization process to encapsulate functional species into MOF crystals. The reversible phase transition avoids high temperature, high ionic strength, strong acid/base conditions, etc., which is suitable for many types of functional species. To prove the feasibility of this method, enzymes, anti-cancer drugs, noble metal nanoparticles and other functional materials, have been trapped into MOFs using this strategy. The synthesized MOF composites can maintain 95.6% of the enzyme activity under the treatment of protease, or reach 40% drug loading, or achieve 98% size selectivity for olefins. This strategy has been extended to several types of MOF structures and it will pave a new way for designing MOF composites and developing further applications.

during the reversible phase transition of MOFs without involving any harsh conditions, and therefore is applicable to many types of functional materials. By using this strategy, a series of functionalized ZIF-8 composites have been synthesized, which achieved synergistic effects including the protection of enzyme, high drug loading, size selective catalysis, etc. In addition, PASR strategy has been demonstrated to be versatile and extendable to other types of MOFs (such as Co-ZIF-67, Cu-HKUST-1, Al-MIL-53 and Mg-MOF-74) and functional guests (such as enzymes, multi-wall carbon nanotubes (MWCNTs), Au NPs, TiO2 NPs and SiO2 NPs).

PASR for MOFs
As a proof of concept, the procedure of the PASR for pristine MOFs was demonstrated by ZIF-8 nanocrystals. First, 10 mg of activated ZIF-8 was ground and transferred into a 13 mm pellet die, and then certain mechanical pressure (0 MPa-1,364 MPa) was applied. The effect of compression stress on the crystal structure has been systematically studied. The crystal structure, morphology and porosity of ZIF-8 were characterized by powder X-ray diffraction (PXRD), scanning electron microscopy (SEM) and Brunner-Emmet-Teller (BET), respectively (Supplementary Fig. 1, Supplementary Fig. 2 and Supplementary Fig. 5). The characteristic diffraction peaks of ZIF-8 merged into broad peaks at 1,364 MPa (Fig. 1a). Low compression loading only induced partial assembly of ZIF-8 NPs (Fig. 1c), while high compression loading resulted in the global formation of tightly packed aZIF-8 NPs (Fig. 1d). These results are in agreement with previous works on pressure-induced aZIF-8 in diamond anvil cell pressure apparatus 36 and confirm the successful synthesis of aZIF-8 pellets with compression loading of 1,364 MPa.
Conventionally, ZIF-8 is synthesized by dissolving metal ions and organic ligands in methanol solvent. Accordingly, methanol was chosen as the recovery solvent to gradually repair aMOFs. The prepared aZIF-8 pellets (compressed by 1,364 MPa) were individually exposed to saturated methanol vapor for 24 hours or immersed in methanol for 5 hours at 50°C to recover the crystal structure. As expected, the crystallinity of ZIF-8 was reproduced as indicated by PXRD and SEM measurements.
Significant surface morphology change was observed between the amorphized and recrystallized ZIF-8 (Figs. 1e and 1f and Supplementary Figs. 6 and 7). As compared to the amorphized sample, polyhedral nanocrystals with varied sizes were found in both recovered samples. The original ZIF-8 NPs had a specific area of 1,347 m 2 /g and pore volume of 1.28 cm 3 /g·nm. After exposing aZIF-8 (surface area: 406.7 m 2 /g, pore volume: 0.465 cm 3 /g·nm) with methanol vapor and soaking in methanol solution, the specific areas of ZIF-8 pellets increased to 881.6 m 2 /g and 1,027.9 m 2 /g ( Fig.   1b), respectively, and the pore volumes were slightly increased to 0.62 cm 3 /g·nm and 0.77 cm 3 /g·nm, respectively ( Supplementary Fig. 8). The above experimental results jointly prove that the pressure-induced aZIF-8 can be mostly restored to the crystalline ZIF-8 under the stimulation of either methanol vapor or methanol solvents.
Accordingly, water has also been used to recover aZIF-8 41 , but we found that it may induce the aZIF-8 to transform into non-porous diamondoid (dia)-ZIF and ZIF-CO3-1 structures 42 , as reflected by PXRD (Supplementary Fig. 9) and SEM images ( Supplementary Fig. 10). It is clear that two new morphologies different from ZIF-8 crystals have appeared after soaking aZIF-8 in water for 72 hours.
To further accelerate the recovery of aZIF-8 into the crystalline one, 2-methyl imidazole/aqueous solvents were utilized because the 2-methyl imidazole could stabilize the structures of ZIF-8 and avoid further phase transition of ZIF-8 into other ZIF structures (Supplementary Fig. 9 and 10).
The PASR strategy is not limited to ZIF-8, but can be further extended to other MOFs (Co-ZIF-67, Cu-HKUST-1 Mg-MOF-74 and Al-MIL-53) ( Supplementary Fig. 11), bringing MOFs broader application prospects. PXRD results demonstrated that all these MOFs could be amorphized by compression and subsequently reconstructed by methanol (or water) vapor/solvents exposure ( Supplementary Fig. 12), indicating the proposed PASR strategy is suitable for certain MOFs.

PASR for the encapsulation of enzymes in ZIF-8
In order to verify the feasibility of PASR strategy in encapsulation, we tried to incorporate glucose oxidase (GOx) and horseradish peroxidase (HRP) into ZIF-8. After applying an axial pressure loading of 1,364 MPa to 10 mg GOx&HRP for 5 min, the enzymes' activity remained almost the same as free enzymes (Fig. 2i). This implied that such pressure will not affect the activity of GOx and HRP. Based on cascade reactions for GOx and HRP 32 , 90 mg ZIF-8, 5 mg GOx and 5 mg HRP were fully ground and mixed in a mortar, and then 20 mg of the mixture were pressed and recovered in methanol vapor (MV), water vapor (WV) or a trace amount of dimethylimidazole/water (MW) to obtain restored products, GOx&HRP/ZIF-8-X (X = MV, WV or MW). GOx&HRP/ZIF-8-X all showed the characteristic Bragg diffraction peaks consistent with commercial ZIF-8 NPs (Fig. 2a), but few new crystal structures such as dia-ZIF and ZIF-CO3-1 inevitably appeared during the recovery process of GOx&HRP/ZIF-8-WV and GOx&HRP/ZIF-8-MW. From the N2 adsorption results ( Supplementary Fig. 13), we found GOx&HRP/ZIF-8-MV had a relatively high specific surface area, indicating that the conversion degree from aZIF-8 to crystalline ZIF-8 was the highest in these samples. Through the pore size distribution, it was observed that the porosity of all the prepared GOx&HRP/ZIF-8-X was recovered (Fig. 2b). Notably, a small amount of mesopores appeared in GOx&HRP/ZIF-8-WV, suggesting some defects of ZIF-8 were induced during phase transformation (Fig. 2b)  The most important purpose of preparing enzyme/MOF composites is to use MOF matrix to protect the biological activity of enzymes. To prove that the enzymes encapsulated in MOFs have similar activity as free enzymes, the biological activity of GOx&HRP/ZIF-8-X were studied in tandem reactions. During the reactions, GOx converted glucose into gluconic acid and generated H2O2, which is the substrate for HRP to oxidize 2,2'-diazo-bis-3-ethyl benzothiazoline-6-sulfonic acid (ABTS 2-) to ABTS -• . The absorbance at 415 nm of ABTS -• is monitored as a standard for reaction progress.
The activities of GOx&HRP/ZIF-8-MV and GOx&HRP/ZIF-8-MW were preserved (55.1%, 89.1%), but lower than the activity of free enzymes (Fig. 2i), implying that organic solvents and organic salts had negative effects on the activity of enzymes 33,[43][44][45] . In addition, the small windows of ZIF-8 (3.4 Å) would restrict the transportation of substrate, which might be accountable for the lower activities as well. Interestingly, due to the presence of mesopores, GOx&HRP/ZIF-8-WV exhibited the highest activity (99.8%), which favored the acceleration of reaction rates and the contact between the substrate and enzyme 46,47 . The contents of the two labeled enzymes in ZIF-8 were determined by using a fluorescence standard curve. GOx&HRP/aZIF-8 had 3.96% of GOx and 1.09% of HRP, probably due to the leakage of the free enzyme from aZIF-8 NPs, resulting in its lower loading amount. The GOx&HRP/ZIF-8-WV contained 4.98% of GOx and 1.78% of HRP, indicating that the recovery process is beneficial to the encapsulation of the enzyme ( Supplementary Fig. 18). The experiment of encapsulating one single enzyme was also carried out and proved that the enzyme activity (94.7% for GOx and 82.4% for HRP) was maintained under the condition of water vapor recovery ( Supplementary Fig. 19). These results prove that PASR strategy can maintain the activity of enzyme after encapsulation.
To investigate whether the ZIF-8 structure could keep the enzyme from being attacked by other inhibitors, the free enzyme, GOx&HRP/aZIF-8 and GOx&HRP/ZIF-8-WV were immersed in 1 mg/mL protease and 1 wt% ethylene diamine tetraacetic acid (EDTA) solution for 1 hour, respectively. The GOx&HRP/ZIF-8-WV maintained 95.6% and 92.4% of enzyme activity under the treatment of protease and EDTA, respectively, whereas the activity of GOx&HRP/aZIF-8 decreased to 66.3% and 54.9% of the original value (Fig. 2j). The low loadings of enzymes and poor mass transfer caused the low activity of GOx&HRP/aZIF-8 while the mesopores and micropores from GOx&HRP/ZIF-8-WV accelerated the reaction rate and protected enzymes from being attacked by inhibitors. For free enzymes, the activities they remained were only 28.7% and 38.7%. The results show that PASR strategy can not only encapsulate but also protect enzymes desirably.

PASR for encapsulation of DOX molecules in ZIF-8
After successfully using the PASR encapsulation strategy to prepare GOx&HRP/ZIF-8-WV with high activity and stability, we tried to apply this strategy to prepare more guest/MOF composites, such as drugs/MOF composites. Doxorubicin (DOX) was used as guest molecules to be encapsulated  It is known that some NPs are difficult to be encapsulated into MOFs without the assistance of surfactants such as PVP 18 or using double-solvents approach 51  in MOF structures as observed by SEM mapping and TEM images (Fig. 4, Supplementary Fig. 33).
The catalytic selectivity of Pt/ZIF-8-MS for 1-hexene molecules was increased to 98%, indicating the encapsulation of Pt NPs in Pt/ZIF-8-MS (Supplementary Fig.31). In general, the PASR method can encapsulate different types of guest species into MOFs with different structures, which illustrates the feasibility and versatility of the PASR strategy for preparing guest/MOF composites.

Discussion
The PASR shows its specialty on three key factors. One is the synthesis of MOF composites.
GOx&HRP/ZIF-8-WV was prepared by exposing to saturated water vapor conditions at 25°C for 72 hours to fully restore their structure in a vacuum drier at Aseptic carto. GOx&HRP/ZIF-8-MW was prepared by a drop of 2-methyl imidazole/water at room temperature overnight. After recovery, the samples were soaked in 1 mL of deionized water to remove free enzymes and aggregates from the surface. The dry powders were subjected to average pressures of 1,364 MPa (9-ton, 13 mm diameter pellet die) for 5 min. GOx/ZIF-8-MV was prepared by exposing GOx/aZIF-8 to methanol vapor at room temperature for 24 hours. GOx/ZIF-8-WV was prepared by exposing to saturated water vapor conditions at 25°C for 72 hours to fully restore their structure in a vacuum drier. GOx/ZIF-8-MW was prepared by a drop of 2-methyl imidazole/water at room temperature overnight. After recovery, the samples were soaked in 1 mL of deionized water to remove free enzymes and aggregates from the surface.

Preparation of pressed GOx/aZIF-8, HRP/aZIF-8 composite and recovered
The experiments for encapsulating HRP were the same as those for encapsulating GOx except that For the enzymatic activity of free enzymes, the experiments were the same as those above except that the amounts of free enzymes were 1 mg glucose oxidase and 0.2 mg horseradish peroxidase. For the enzymatic activity of free enzymes, the experiments were the same as those above except that the amounts of free enzymes were 1 mg glucose oxidase and 0.2 mg horseradish peroxidase.

Enzymatic activity of recovered GOx/ZIF-8 and HRP/ZIF
Procedures for labeling enzymes with dyes 32 . According to previous report, 8 mg of FITC or N-Succinimidyl 7-hydroxycoumarin-3-carboxylate dissolved in DMSO (2 mg/mL) was slowly added in to 1 mL of GOx or HRP solution (5 mg/mL of enzyme in 0.5 M, pH 9.5 carbonate buffer).
The solution was shaken for 6 h at 150 rpm at 37°C in dark. Free FITC or coumarin was removed via dialysis against de-ionized water. The fluorescent molecule-labeled enzymes were freeze-dried and then dissolved in water for the subsequent synthesis of fluorescently labeled GOx&HRP/ZIF-8 composite. Laser scanning confocal microscope images were taken on a Zeiss LSM880 NLO (2 + 1 with BIG) confocal microscope. The detection wavelengths were 488 nm and 405 nm for FITC and coumarin, respectively.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.