Louis
Longley
a,
Courtney
Calahoo
b,
Thomas J. F.
Southern
a,
Rachel C.
Evans
a,
Lothar
Wondraczek
b and
Thomas D.
Bennett
*a
aDepartment of Materials Science and Metallurgy, University of Cambridge, CB3 0FS, UK. E-mail: tdb35@cam.ac.uk
bOtto Schott Institute of Materials Research, University of Jena, Fraunhoferstrasse 6, 07743 Jena, Germany
First published on 9th February 2021
The thermal behaviour of ZIF-8, Zn(meIm)2 in the presence of a sodium fluoroaluminophosphate glass melt was probed through differential scanning calorimetry and thermogravimetric analysis. The structural integrity of ZIF-8 was then determined by a combination of powder X-ray diffraction, Fourier transform infra-red and 1H nuclear magnetic resonance spectroscopy.
The chemical, mechanical, and thermal stabilities of functional MOF structures are also important considerations in their in-service use. Zeolitic imidazolate frameworks (ZIFs) are a subset of MOFs comprised of divalent metal cations coordinated tetrahedrally by imidazolate linkers,14 whose stability has been extensively studied.15 Advantages of ZIFs include their thermal stability in a wide variety of atmospheres, though care needs to be taken given their reported acid sensitivity and weak mechanical stability under ball-milling.16,17 Further problems arise when considering stability during composite formation, such as reported correlations between incorporated ZIF amount, and mechanical failure in MMMs.18,19
Motivated by the promising results obtained when using a ZIF glass matrix to embed a variety of different crystalline MOFs into composite materials,11,12,20 this work examines the possibility of expanding the MOF CGC concept, to incorporate structures in which the glass matrix is an inorganic glass. This would greatly expand the range of different matrices available, as the number of glass forming MOFs is extremely limited compared to the vast amount of known inorganic glasses. In order to form a successful MOF-inorganic CGC two criteria must be fulfilled: (i) Thermal compatibility; a processing temperature must be found at which the inorganic glass matrix is sufficiently above its glass transition (Tg), that its viscosity is low enough to promote mixing and composite formation. The temperature however, should also be low enough that the crystalline MOF remains thermally stable. (ii) Chemical compatibility; the MOF crystal inorganic glass pair must have compatible chemical bonding such that they can be bonded at the interface to produce a robust material, and no reactions which promote decomposition should occur during high temperature liquid phase mixing between the MOF and inorganic components.
Previous work has demonstrated successful composite formation between ZIF-62 (Zn(Im)1.75(bIm)0.25 Im = imidazolate C3H3N2−, bIm = benzimidazolate C7H5N2−) glass and a series of fluoroaluminophosphate glasses, where the final composite material formed was composed of separate inorganic glass and ZIF-62 glass domains which displayed bonding at the interface.21 That work therefore indicates that these glasses are chemically compatible with ZIF-62 and therefore may imply a more general chemical compatibility with other ZIF frameworks. The lowest Tg glass of the series, 78([Na2O]1.6[P2O5])-22([AlO1.5][AlF3]0.7), was selected for use as a matrix material in this study because it would enable the use of the lowest processing temperatures and therefore increase the likelihood of the crystalline MOF being thermally compatible. ZIF-8, Zn(meIm)2 (meIm = 2-methylimidazolate, C4H5N2−), was chosen as it is the ‘prototypical’ crystalline ZIF, whose properties and structure have been extensively studied and which is available commercially under the name Basolite® Z1200. ZIF-8 is a monolinker framework (Fig. 1a) with the cubic space group I3m and the sod topology (Fig. 1b).22,23 The local structure of the inorganic glass is composed of unconstrained tetrahedrally coordinated phosphate chains, which are strongly bonded to highly constrained, octahedrally coordinated aluminiumoxyfluoride islands, Al(OP)4F2 (Fig. 1c).24
![]() | ||
Fig. 1 (a) Local structure around the zinc centre in ZIF-8. (b) Unit cell of ZIF-822 viewed down the b axis. Element key. N: blue, C: grey, Zn: green, H omitted for clarity. (c) Schematic of the local bonding around the phosphorous and aluminium polyhedra in the inorganic glass, m indicates phosphate tetrahedra of varying polymerisation (chains, branching, etc.). The degree of localisation of the PO double bond depends on the local P environment. |
The thermal responses of the samples were measured during heating at 10 °C min−1 to 600 °C under argon (Table S1 and Fig. S3 ESI†). The differential scanning calorimetry (DSC) heating curve of the inorganic glass control was flat until a glass transition (Tg) with an onset at 350 °C, which was followed by an exotherm, with an onset of 530 °C, ascribed to recrystallisation. The heating curve of a ZIF-8 control sample was relatively featureless until around 500 °C, then became rapidly endothermic, in accordance with thermal decomposition (Fig. S3 ESI†). The 5 wt%, 10 wt% and 15 wt% ZIF-8 samples displayed similar behaviour to the inorganic glass, with Tgs between 350–360 °C followed by exotherms with onsets at approx. 520 °C (Table S1 ESI†). The recrystallisation exotherm in the 30 wt% ZIF-8 sample was much shallower and harder to unambiguously assign, this is ascribed to the increased ZIF-8 content.
The mass curves of the controls and samples were also measured by thermogravimetric analysis (TGA) (Fig. 2). The mass curve of the inorganic glass control showed a small loss of approx. 1% at 600 °C. A small peak in the mass curve was observed at approx. 540 °C, coinciding with the exotherm in the DSC curve and attributed to an increase in noise during recrystallisation. The ZIF-8 control lost approximately 3% of its mass before 500 °C, indicating successful activation of the framework. Above 500 °C, the mass loss increases sharply, in line with the rise in the DSC baseline and further confirming this as the onset of decomposition (TZIF-8d). The 5 wt% and 10 wt% mixtures have mass curves that show intermediate behaviour between the controls, as would be expected from a rule of mixtures, however after 400 °C the 15 wt% and 30 wt% samples both lost mass at a faster rate than the ZIF-8 control.
![]() | ||
Fig. 2 Thermogravimetric analysis (TGA) curves of the ZIF-8, inorganic glass and composite samples in the region 250–550 °C. Inset: the same TGA data shown over the full measured range. |
In order to examine this unexpected effect the following construction was used; the average mass percentage signals measured from the inorganic glass control, mInorg, and ZIF-8 control, mZIF (Table S1 ESI†) were combined in the appropriate mass proportions, xZIF and xInorg respectively, as a measure of the mass loss that would be expected from a non-interacting mechanical mixture of the ZIF-8 and inorganic (m(T)non-int):
m(T)non-int = xZIFm(T)ZIF + xInorgm(T)Inorg | (1) |
This can then be subtracted from the mass signal directly measured for each of the mixtures in order to approximate the mass loss due to interaction between the inorganic glass and ZIF-8 (Δm(T)):
Δm(T) = m(T)mixture − m(T)non-int | (2) |
The results of this calculation (Fig. 3) reveal that below 300 °C all Δm(T) values are near zero, indicating relatively little interaction between the ZIF-8 and inorganic glass powder. However, at elevated temperatures, all samples lose more mass than would be expected from a non-interacting mixture (Δm < 0). Although Δm(T) is consistently negative above 300 °C, at T > 400 °C the gradient increases rapidly. Above 520 °C, i.e. in the region in which the ZIF-8 control decomposes on its own, Δm(T) begins to level off.
To further investigate the origin of this effect, the X wt% ZIF-8 samples were compressed into pellets to promote production of mechanically robust composites (see Experimental). Heat treatments of 450 °C for 30 minutes and 480 °C for 1 minute were selected as a compromise between promoting flow in the inorganic glass to join the ZIF-8 and glass together into a solid body, while avoiding decomposition of the ZIF-8 and recrystallisation of the inorganic glass. These samples were produced by heating in a DSC-TGA experiment (see Experimental).
PXRD was then conducted on the X wt% ZIF-8 composites, as well as the ZIF-8 and the inorganic glass controls before and after heat treatment (Fig. S4–S9 ESI†). The area under the three most intense peaks of the ZIF-8, at approximately 7.3, 12.7 and 18.0° 2θ, was used as metric for the relative change in crystallinity across the series (see section 1 in ESI†), and expressed as a fractional percentage of the peak area in the evacuated, ball-milled ZIF-8. The results show that some crystallinity is lost on pelletisation in the case of the ZIF-8 control and the 30 wt% ZIF-8 sample, but relatively little change is observed in the 5, 10 and 15 wt% ZIF-8 samples (Fig. 4, Table S2 ESI†). This difference is ascribed to the effect of the relatively dense and rigid inorganic glass, which shields the ZIF-8 from some applied stress. The 30 wt% sample displays far less stress-shielding than the other samples due to its higher fraction of ZIF-8; estimated as being over 50% by volume using densities reported in the literature.21,31
![]() | ||
Fig. 4 Change in crystallinity, relative to the ball-milled evacuated ZIF-8 control. Inset: the heat treated and pellet pressed samples. |
These results also show that while the crystallinity of the ZIF-8 control did not decrease further on heating, the crystallinity of all the X wt% ZIF-8 samples was reduced on heat treatment, with some samples appearing completely amorphous after heat treatment (Fig. 4, Table S2 ESI†).
ZIF-8 decomposition at elevated temperatures and in different atmospheres has previously been studied by FTIR spectroscopy.32 FTIR of the ZIF-8 control samples contained peaks matching those recorded in the literature for as-synthesised ZIF-8, except for a broad peak at approximately 3000 cm−1, due to O–H bonding, the absence of which in these samples is attributed to framework activation (Fig. S10 ESI†). The FTIR spectra of the ZIF-8 controls were unchanged upon heat treatment (Fig. S10 ESI†). The inorganic glass controls showed very broad absorbances (Fig. S11 ESI†), which is typical of IR spectroscopy on glasses where the high degree of structural disorder leads to a wide variety of similar vibrational modes.33 The positions of these peaks were found to match those reported for other polymeric phosphates (Fig. S11 ESI†).33,34 The heat-treated and pellet pressed inorganic glass controls also showed very little change. A band at approximately 2300 cm−1, which is observed in all samples and controls, though with very variable intensity, was ascribed to atmospheric CO2 present in the open measurement set-up.35
In order to investigate the changes in the relative proportions of the inorganic glass and ZIF-8 in the samples during heat treatment, the area under the ZIF-8 peak at 1440 cm−1 and the inorganic glass peak at 910 cm−1 were integrated and a ratio taken (see section 2 in ESI†). These ratios show a general decrease in ZIF-8 content on heat treatment (Fig. S12–23, Table S3 ESI†). Peaks at 904 cm−1, 1041 cm−1, 1251 cm−1 and 2200 cm−1 are reported in the literature to appear upon decomposition, and are ascribed to disordering in the CN bonding and the formation of aliphatic amines.32 Some of these peaks are indeed observed in the 10 wt%, 15 wt% and 30 wt% samples on heat treatment, though due to the inorganic glass background their unambiguous assignment is difficult. They are not observed in the 5 wt% sample, which is due to the overall low ZIF-8 content (Fig. S12–23 ESI†).
To confirm that the reduction in Bragg scattering in PXRD, and relative ZIF-8 peak intensity in FTIR coincides with a reduction in the amount of 2-methylimidazole, 1H NMR spectroscopy was conducted on the 15 wt% ZIF-8 sample heated to 450 °C for 30 minutes (Fig. S24–28, Tables S4 and S5 ESI†). This sample was selected because it showed a substantial drop in ZIF-8 content, as measured by PXRD and FTIR. Although peaks matching those of the ZIF-8 control were observed, their integration trace had fallen by a factor of five relative to the non-heat-treated sample, using the DMSO-d6 peak as a standard (Fig. S24–28, Tables S4 and S5 ESI†). This confirms the results of the FTIR experiments, showing that the 2-methylimidazolate linker is destroyed during heat treatment, and indicates that the reduction in the intensity of the Bragg peaks is due to decomposition of the linker and not to thermal amorphisation of the framework. However, no decomposition products could be observed in the 1HNMR results. This is explained by decomposition being reported as proceeding via formation of tertiary amines with no H environments.32
The results presented here indicate that an interaction occurs between ZIF-8 and the inorganic glass, which has a destabilising effect on the ZIF framework. The fact that Δm(T) becomes rapidly more negative at temperatures above 400 °C, i.e. approx. 50 °C above TInorgg, is consistent with increasing decomposition due to a more intimate mixing between the inorganic and the ZIF, i.e. mixing occurring when the inorganic glass is sufficiently fluid. PXRD data indicate the reduction or complete loss of crystallinity on heating the pellet pressed X wt% samples. FTIR and 1H NMR confirm that this loss is due to linker decomposition rather than amorphisation of ZIF-8. New peaks that emerged during heat-treatment in the FTIR spectra are consistent with those previously reported as being observed during decomposition of ZIF-8 in the literature.32 This indicates that decomposition in the inorganic ZIF-8 mixtures may proceed via a similar mechanism to decomposition of the pure ZIF-8, but that it occurs at a lower temperature due to the presence of the inorganic glass.
Previous work has shown that ZIF-62 formed stable composites with this inorganic glass and so the instability of ZIF-8, a framework with a similar chemistry, is somewhat surprising.21 SEM on evacuated and dispersed samples of both frameworks coupled with image processing and analysis (see section 4 ESI†) confirmed that this difference was not due to a drastically smaller particle size in the ZIF-8 relative to the ZIF-62, with the ZIF-8 in this study having a comparable range of particle sizes to the ZIF-62 used previously (Fig. S29–S34 ESI†).
This leaves differences which are intrinsic to the two frameworks, namely linker chemistry and framework topology, which have both been found to be key factors in determining framework stability.36 More dense frameworks, for example, are reported as being more stable due to enhanced dispersion interactions between the linkers,37 and so the instability of ZIF-8 relative to ZIF-62 could therefore be ascribed to its lower framework density.22,38 The addition of a methyl group to the imidazole ring has also been shown by molecular dynamics simulations to have a substantial effect on the high-temperature dynamics around the Zn centre39 and so will also have an effect on the linker reactivity.
These findings illustrate the difficulty of finding chemically compatible MOF-inorganic glass pairs. Interfacial interactions are inherently difficult to characterise due to the small size of interfacial regions relative to the bulk of both components in the composite. This means that interface specific interactions, though vital to successful composite formation, are often not observed by bulk experimental methods. Moreover, interactions at the interfaces involving decomposition may be even more difficult to fully characterise due to the transient nature of the decomposing ZIF framework which makes ex situ measurement difficult. The results here underscore the inherent difficulty of predicting chemical compatibility ahead of time.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1dt00152c |
This journal is © The Royal Society of Chemistry 2021 |