Formation of a meltable purinate metal–organic framework and its glass analogue

The chemistries that can be incorporated within melt-quenched zeolitic imidazolate framework (ZIF) glasses are currently limited. Here we describe the preparation of a previously unknown purine-containing ZIF which we name ZIF-UC-7. We find that it melts and forms a glass at one of the lowest temperatures reported for 3D hybrid frameworks.

Metal-organic frameworks (MOFs) are a widely studied family of porous materials composed of inorganic nodes and organic molecular linkers. 1,2 Interest in these materials has increased in recent years due to their wide variety of potential applications encompassing heterogeneous catalysis, 3,4 gas storage and separation, 5,6 and biomedicine. 7,8 Zeolitic imidazolate frameworks (ZIFs) are a subset of MOFs composed of tetrahedrally coordinated M 2+ metal nodes connected to imidazolate linkers that are well known for their high thermal and chemical stabilities. 9,10 Recently, it has been demonstrated that certain ZIFs can melt. 11 Quenching these materials from the liquid state results in the preparation of melt-quenched MOF glasses-the first new category of glass reported in 50 years. 12,13 Melting in ZIFs is a dynamic process, involving rapid decoordination and re-coordination of the imidazolate linkers. 11 However, even though more than one hundred ZIF structures have been reported, 14 only a handful of ZIFs have been found to form melt-quenched glasses. 12,13 The most well-studied glass-forming ZIF is ZIF-62 [Zn(Im) 2Àx (bIm) x ] (Im -imidazolate -C 3 H 3 N 2 À ) (bIm -benzimidazolate -C 7 H 5 N 2 À ) which crystallises in the Pbca space group and displays the cag network topology. [15][16][17] Various studies have demonstrated that it is possible to alter the melting behaviour of ZIF-62 by changing the metal cation from Zn 2+ to Co 2+ , as well as by altering the ratio of Im to bIm in the structure. [18][19][20] Other cag topology ZIFs have since been shown to melt, with different functionalised bIm linkers incorporated to alter their thermal behaviour. [21][22][23] However, the incorporation of other organic linker chemistries into glass-forming ZIFs has yet to be reported. The use of MOFs in biomedical applications is gaining increasing interest due to their tuneable pore sizes and varied chemistry. 7 For these applications, it is essential that MOFs contain biocompatible metal ions such as Zr 4+ , Zn 2+ or Fe 2+ , 8,24 and biocompatible organic linkers such as amino acids, peptides, carbohydrates and nucleobases. [25][26][27] Various MOFs have been prepared using the nucleobase adenine as an organic linker and these frameworks have shown promise for CO 2 uptake, heterogeneous catalysis and drug delivery. [28][29][30] Adenine is part of the family of molecules known as purines, the simplest of these being purine itself. Purine has already been incorporated as an organic component in various ZIFs, [31][32][33] however, the thermal behaviour of purine-containing ZIFs has not been explored in detail.
It is not only MOFs that have been proposed for biomedical applications. Inorganic glasses, including phosphates and borates, are promising candidates for biomedical applications including drug delivery and bone tissue regeneration. 34,35 Given the promise of both crystalline MOFs and inorganic glasses in biomedicine, the motivation of this study was to see how biological functionality may be incorporated into a MOF glass. Here we describe the preparation of a previously unknown purine-containing ZIF, denoted ZIF-UC-7 ( Fig. 1), before demonstrating its melting and glass-forming behaviour.
Solvothermal synthesis was used to prepare orange single crystals of ZIF-UC-7 (Methods and Fig. S1, S2, ESI †). Specifically, zinc nitrate hexahydrate (5.09 mmol), imidazole (108 mmol) and purine (12. Table S1, ESI †). ZIF-UC-7 therefore crystallises with a very similar unit cell to many liquid and glassforming ZIFs. 18 The thermal behaviour of ZIF-UC-7 was then studied in detail (Fig. 2). Thermogravimetric analysis (TGA) revealed that thermal decomposition began above T d = 550 1C (Fig. S7, ESI †). Differential scanning calorimetry (DSC) was then used to locate a melting endotherm (T m = 318 1C) which was lower than the onset of thermal decomposition (Fig. S8, ESI †). However, these measurements displayed a second, broad endotherm which obscured the melting event (Fig. S8, ESI †). Assessment of the TGA trace revealed that melting was accompanied by a ca. 3% mass loss (Fig. S7, ESI †). Furthermore, solution state 1 H NMR spectroscopy revealed that DMF was present in the ZIF (Fig. S6, ESI †), despite the rigorous washing and activation procedure used (Methods). The broad feature in the DSC is therefore attributed to the release of DMF from the framework during heating. 1 H NMR spectroscopy and CHN microanalysis of a sample of ZIF-UC-7 heated to 400 1C then confirmed this DMF can be thermally removed ( Fig. S9 and Tables S3, S4, ESI †).
An in situ activation procedure was then performed in the DSC instrument to isolate the melting endotherm (Fig. S10, ESI †). First, a sample of ZIF-UC-7 was held at 275 1C under an argon atmosphere for five minutes to remove excess DMF from the structure ( Fig. S11 and Table S5, ESI †). This pre-heated sample was cooled to 30 1C before heating to 350 1C to observe the unobscured melting event (T m = 318 1C) ( Fig. 2 and Fig. S12, ESI †). Finally, this sample was cooled to 30 1C before heating a third time to locate its glass transition (T g = 273 1C) (Fig. 2 and Fig. S12, ESI †), i.e., the temperature where the glass transforms to a more flowable, liquid-like state. Hereafter, samples of ZIF-UC-7 melt-quenched from above T m will be denoted a g ZIF-UC-7.
1 H NMR spectroscopy confirmed that a g ZIF-UC-7 has the same composition as crystalline ZIF-UC-7, i.e., [Zn(Im) 1.75 (pur) 0.25 ], with no decomposition observed until 340 1C (Fig. S13, ESI †). Moreover, only diffuse scattering was observed in its X-ray diffraction pattern (Fig. S3, ESI †), supporting its transformation to the glass phase. Furthermore, particle coalescence and sample flow were evident from optical microscopy and scanning electron microscopy (SEM) (Fig. 2, 3 and Fig. S15, ESI †). The melting event was then investigated further by heating separate samples of ZIF-UC-7 to 250 1C, 275 1C, 305 1C, 315 1C, 350 1C, 400 1C and 450 1C followed by ex situ powder X-ray diffraction (PXRD) (Fig. S16, ESI †). Bragg reflections from ZIF-UC-7   were observed for samples heated to temperatures below 305 1C. However, for samples heated above this temperature, all reflections were lost, and only diffuse scattering was observed. ZIF-UC-7 therefore displays a large working temperature range for a ZIF (i.e., the temperature range between T m and T d , Table S7). 12 The structure of a g ZIF-UC-7 was then compared to that of crystalline ZIF-UC-7 using X-ray total scattering data collected at beamline I15-1 at the Diamond Light Source (Fig. S17, ESI †). Fourier transformation of the corrected total scattering data yielded their pair distribution functions (PDFs) (Fig. 4). The local structure of a g ZIF-UC-7 was similar to that of ZIF-UC-7. Both materials displayed the same short-range order correlations (labelled 1-5). Correlations 1 to 5 were dominated by CQC/C-N, Zn-N, ZnÁÁÁC, ZnÁÁÁN and ZnÁÁÁZn distances, in turn, which is very similar to ZIF-62. 37 However, the change in linker chemistry from bIm to pur (i.e., the replacement of two C atoms with N atoms) will lead to a marginal increase in weighting for all N-containing partial PDFs within the total PDF. Furthermore, the static disorder for both the Im and pur linkers will result in slight peak broadening due to the presence of different configurations. Nevertheless, the lack of distinct correlations over longer distances in a g ZIF-UC-7 confirms its transformation to the glass phase. CO 2 sorption isotherms were then collected at 273 K for both crystalline ZIF-UC-7 and a g ZIF-UC-7 (Fig. 5). ZIF-UC-7 has a maximum CO 2 uptake of 48.1 cm 3 g À1 (Fig. S18 and Table S8, ESI †). This is comparable to the CO 2 uptake reported for other glass-forming ZIFs 21,22 and approaches some of the highest values of CO 2 uptake reported to date with a cag topology (Table S9, ESI †). 10,38 This high CO 2 uptake is likely due to the large number of nitrogen atoms in the framework from the purinate linker, as nitrogen atoms have previously been shown to have a high affinity for Lewis acidic gases such as CO 2 . 5 As expected, a g ZIF-UC-7 has a lower CO 2 uptake (18.9 cm 3 g À1 ) ( Fig. S19 and Table S10, ESI †) due to the pore network collapsing upon glass formation. This value is similar to the uptake values of other glassy ZIFs. 22 However, although glass formation results in a reduction in CO 2 uptake, it does not completely preclude framework porosity.
ZIF-UC-7 displays one of the lowest T m values recorded for a ZIF. 13 To rationalise this behaviour, the enthalpic (DH fus ) and entropic (DS fus ) contributions to melting were calculated based on reported methods 39 and compared to values calculated for other ZIFs (Fig. S20 and Table S7, ESI †). 22,36 Of all these systems, ZIF-UC-7 had the lowest value for DH fus , where lower DH fus values result in a reduced T m . 39 Therefore, as has been found for other ZIFs, 22 the low T m of ZIF-UC-7 may result from an enthalpic effect, such as weaker Zn-N coordination bonds in the framework. However, other factors may also be contributing to the low T m of ZIF-UC-7. For example, during melting the uncoordinated, recently detached, purinate ion may be more stable than an equivalent benzimidazolate ion, as the negative charge can be delocalised across four nitrogen atoms in the purinate ion. A more stable uncoordinated linker could potentially lower the energy barrier for the dissociative melting mechanism, resulting in a reduction in T m . In addition, the purinate linker contains additional N donor atoms. These may act as additional transient coordination sites during melting, helping to stabilise the uncoordinated linkers that form during the melting process. Regardless of the factors contributing to melting, the T m of ZIF-UC-7 is remarkably low compared to other ZIFs, which may be advantageous for material processing.
The T g for a g ZIF-UC-7 (T g = 273 1C) is also one of the lowest values reported for a ZIF. 13,20 This may also be attributed to an enthalpic effect, such as weaker coordination bonds in the  framework. 39 Additionally, the smaller size of the purinate linker (steric index ca. 618 Å 4 ) compared to a substituted benzimidazolate linker (steric index ca. 915 Å 4 ) may result in an increase in steric freedom in the glass and hence a reduction in T g . 12,40,41 The use of purinate linkers in ZIFs and their glasses is therefore a promising strategy to lower both T m and T g .
To conclude, in this work we have expanded the types of possible molecules that can be incorporated within MOF glasses by the successful inclusion of purine in ZIF-UC-7. Moreover, the presence of the purinate linker in ZIF-UC-7 results in ZIF-UC-7 displaying one of the lowest melting temperatures (T m = 318 1C) and glass transition temperatures (T g = 273 1C) reported for any ZIF. Furthermore, its successful incorporation in a ZIF glass provides a promising case study, illustrating how biologically active purines, including nucleobases such as adenine and guanine, may be incorporated within ZIF glasses. A

Conflicts of interest
There are no conflicts to declare.