Phase Transitions in Zeolitic Imidazolate Framework 7: The Importance of Framework Flexibility and Guest-Induced Instability

Zeolite-related materials exhibit a range of novel properties and are of considerable interest for their potential engineering applications. Zeolitic imidazolate frameworks (ZIFs) display zeolite-type structures and are constructed by transitional metals and imidazole molecules.(1) With a wide variety of potential organic ligands, ZIFs present a new family of possible zeolite-related structures with tunable and functionalizable properties. Because of the coordinative metal–imidazolate bonding forming their frameworks, ZIFs are commonly more flexible than their aluminosilicate analogues. They also show unusual gas sorption capacity and related properties. Due to their framework flexibility, ZIFs can undergo structural transformations, e.g., during the sorption process(2) or under high temperature(3) or pressure.(4) It is of great significance to understand potential structural phase transitions since they strongly affect ZIFs’ structurally-related sorption and mechanical properties, which are essential to ZIFs-based technology innovations and industrial applications.Zeolite-related materials exhibit a range of novel properties and are of considerable interest for their potential engineering applications. Zeolitic imidazolate frameworks (ZIFs) display zeolite-type structures and are constructed by transitional metals and imidazole molecules.(1) With a wide variety of potential organic ligands, ZIFs present a new family of possible zeolite-related structures with tunable and functionalizable properties. Because of the coordinative metal–imidazolate bonding forming their frameworks, ZIFs are commonly more flexible than their aluminosilicate analogues. They also show unusual gas sorption capacity and related properties. Due to their framework flexibility, ZIFs can undergo structural transformations, e.g., during the sorption process(2) or under high temperature(3) or pressure.(4) It is of great significance to understand potential structural phase transitions since they strongly affect ZIFs’ structurally-related sorption and mechanical properties, which are essential to ZIFs-based technology innovations and industrial applications.


/ 36
height amplifier discrimination. The generator was operated at 40 kV and 40 mA. The sample was prepared for analysis by gently grinding the product obtained in an agate mortar and then depositing on a flat-plate sample holder using ethanol. Diffraction data were collected at room temperature in the range of 6-50° (2θ), in θ:2θ mode and step-scan with ∆2θ=0.02°, for 2 seconds per step. The XRPD pattern of ZIF-7-I was compared with the simulated pattern from the model given by Yaghi et al., 2006. 2 The amount of zinc oxide (ZnO) impurity is negligible, probably less than 1 % wt.

ZIF-7-I to II phase transition observed by Raman spectroscopy
Raman spectra of an as-synthesized ZIF-7-I sample were collected using a Labram 300 spectrometer (Horiba Jobin-Yvon TM ) of 300 mm focal length equipped with an 1800 grooves/mm grating and a 1024×256 pixels Peltier cooled CCD detector. The excitation radiation at 632.8 nm was produced by an internal HeNe 20mW laser. Before the experiment, a piece of silicon single crystal was used for calibration. ZIF-7-I sample was placed in a Linkam 4 / 36 TH1500 heating stage under an Olympus TM 50× objective (0.5 N.A.). The temperature of the sample was monitored by a K-type thermocouple placed as close as possible to the sample. In the study, the temperature ranged from 297 to 421 K and was increased by 1 K/min. Raman spectra were collected every 10 K. Typical accumulation time was 60 seconds for each spectrum. Raman spectra were fitted using PeakFit 4.12 in sections; baseline was corrected and the peak features were determined with Voigt function.

ZIF-7-I and II phase transitions observed by X-ray powder diffraction
High-temperature XRPD study was carried out on an as-synthesized ZIF-7-I sample using Bruker D8 Advance X-ray diffractometer equipped with a VÅNTEC-1 detector and Goeble mirror for parallel beam optics. The generator was operated at 40 kV and 40 mA. The product obtained from solvothermal synthesis was re-homogenized by gently grinding in an agate mortar and then carefully deposited in an alumina sample holder. The sample chamber was evacuated to 10 -1 Pa before analysis using a dynamic vacuum. The diffraction data were collected using CuKα radiation (λ=1.5418 Å) between 6-45° (2θ) in θ:2θ mode and step-scan with ∆2θ=0.02°, for 1 second per step, initially at 300 K and then in temperature steps of 5 K from 300 to 700 K. The delay time for each step was 60 seconds. The diffraction data were also collected at 610, 520, 430 and 340 K during cooling. The same study was performed on another ZIF-7-I ZIF-7-II 11 / 36 ZIF-7-I sample that had been exchanged with methanol for 48 hours. As the strongest XRPD peaks of ZIF-7-I and II all appear at 2θ below 25° and the peaks of highly crystallized ZnO impurity dominate the diffraction pattern in 2θ above 25°, only raw data in the 2θ range of 6-25° are shown here. The structure of ZIF-7-II product was re-checked by XRPD after being left in air for two weeks. After ZIF-7-II was immersed in DMF for one week at room temperature, its structure was also double-checked by XRPD.       Calorimeter under a nitrogen atmosphere at purge rates of 80 and 150 ml/min. As-synthesized ZIF-7-I sample was sealed into a 40 μl aluminum pan with a pin-hole on the cap, before being placed in the calorimeter. Sample weight was 3.57 mg. The sample was heated from 298 to 573 K at 5 K/min. STARe was used for data acquisition and analysis.

Structure solution of ZIF-7-II using laboratory X-ray powder diffraction
Production: ZIF-7-II sample was produced by heating as-synthesized ZIF-7-I at 400 K in air for 48 hours.

Data collection:
The XRPD data of ZIF-7-II were collected in capillary transmission geometry using a Bruker D8 Advance X-ray diffractometer equipped with a VÅNTEC-1 detector and Goebel mirrors for parallel beam. The generator was operated at 40 kV and 40 mA. The sample was prepared for analysis by gently grinding the product obtained in an agate mortar.
Diffraction data were collected at room temperature using CuKα radiation in the range of 5-70° (2θ), and step-scan with ∆2θ=0.02°, for 12 seconds per step.

/ 36
Structure solution: As the peaks of ZnO impurity dominate the diffraction pattern in 2θ above 30° and all refinable XRPD peaks of ZIF-7-II appear at 2θ below 30°, only raw data in the 2θ range of 5-30° are used for structure solution.
Indexing of the powder pattern using Topas-Academic 4. the loss of one or more symmetry operations from the space group due to the distortion of the structure. 5 Thus, since ZIF-7-I crystallizes in the space group R-3, the likely space groups for the desolvated structure, ZIF-7-II, are P-1, R3 or P-3. As both R3 and P-3 fix the angles of the unit cell at α=β= 90°, γ=120°, these two space groups were discarded from further consideration.
Starting with the reported structure of ZIF-7-I, 2 solvent molecules were removed and the symmetry was reduced to P1. The model was then imported into Material Studio 4.3 6 and the structure was optimized using the Forcite code, allowing all atoms to move under the constraint of fixed connectivity and with the unit cell parameters fixed at the values determined from the Pawley fit. The resulting model was then found to closely match a P-1 symmetry, which was used as the starting point for a Rietveld refinement. 7 The background was described by a 5-term shifted Chebyschev function; two broad Gaussian peaks were necessary to fit the big halo due to the large amorphous fraction. Peak shape was modeled with a pseudo-Voigt function described by CS_L parameter for purely Lorentzian-type crystallite size broadening and Strain_G parameter for microstrain. Spherical harmonics were used for preferred orientation correction. Benzimidazolate ligands were treated as rigid bodies using Cartezian coordinates to reduce the number of variables. Dummy atoms were added to define the origins of rigid bodies. Distance and angle restraints were set between zinc and the coordinating nitrogen atoms with reasonable weight factors. "Anti-bump" restraints were set 23 / 36 between zinc and surrounding carbon and nitrogen atoms with reasonable weight factors.
Single isotropic thermal factors were set for the ligands and zinc element respectively.

Structure refinement of ZIF-7-II using synchrotron X-ray powder diffraction
Production: ZIF-7-II sample was prepared by heating the as-synthesized ZIF-7-I at 450 K under dynamic vacuum for 3 hours. The sample was then exposed to air again for approximately one week before being loaded into a 0.7 mm quartz glass capillary. Glass wool was packed on top of the sample and the capillary then fixed onto a custom made gas cell at beamline I11, Diamond Light Source (Harwell, Didcot, Oxfordshire, United Kingdom). 8 The sample was then heated again to 450 K under dynamic vacuum to ensure complete removal of any aerially adsorbed adsorbate molecules.

Data collection:
The XRPD data of ZIF-7-II were collected in a Debye-Scherrer geometry using X-rays with λ = 0.827142 Å at 300 K. The sample was mounted, as described in a custom built gas cell, which prevented spinning of the sample. Instead the sample was rocked through 20° (-10° to 10°) about the φ axis to improve powder averaging. Data were collected with an array of Mythen II position sensitive detectors over the range 2-90° (2θ), with two data sets offset by 0.25° (2θ) collected (each for 30 seconds). These were subsequently automatically merged to remove gaps in the data where the strips join.
Rietveld refinement: All refinable XRPD peaks of ZIF-7-II appear at 2θ below 30°; only raw data in the 2θ range of 3-21° are used for structure solution. The background was described by an 8term shifted Chebyschev function. Peak shape was modeled by Strain_L parameter; an isotropic model was also used to describe the strain broadening. 9 Spherical harmonics were used for preferred orientation correction. Benzimidazolate ligands were treated as rigid bodies to reduce the number of variables. Dummy atoms were added to define the origins of rigid bodies.
Thermal factors were not refined. Distance restraints were set between zinc and the 24 / 36 coordinating nitrogen atoms. "Anti-bump" restraints were set between zinc and surrounding carbon and nitrogen atoms with reasonable weight factors. Single isotropic thermal factors were set for the ligands and zinc element respectively.

ZIF-7-I to III phase transition observed by X-ray powder diffraction
As-synthesized ZIF-7-I was exchanged with methanol for 48 hours and then water for 48 hours; the product was heated at 400 K for 48 hours. Figure SI 23. The XRPD patterns of as-synthesized, water-exchanged ZIF-7-I, and waterexchanged ZIF-7-I after being heated at 400 K for 48 hours.

ZIF-7-II to III phase transition observed by X-ray powder diffraction
As-produced ZIF-7-II was immersed in water at room temperature for one week.

Synthesis and characterization of ZIF-7-III
Solvothermal synthesis: All chemicals employed were commercially available (Sigma-Aldrich and Acros Organics), with purity of 98 % or above, and were used as received. ZIF-7-III sample was synthesized following a similar synthesis procedure with that of ZIF-7-I: Zn(NO 3 ) 2 •6H 2 O (0.75 g, 2.52 mmol) and HPhIm (0.25 g, 2.05 mmol) were first dissolved in fresh DMF (75 ml).
The resultant solution was then poured and sealed into a 100 ml teflon-lined Parr Bomb. The Parr Bomb was heated at 373 K for 48 hours. After naturally cooling to room temperature, white powdery crystals were isolated after the mother liquor was removed.  Data collection: The ZIF-9-III crystal structure was confirmed by X-ray powder diffraction (XRPD) using Bruker D8 Advance X-ray diffractometer equipped with a Sol-X detector, parallel sollerslits, an incident beam monochromator with CuKα 1 radiation (λ=1.5406 Å) and pulse height amplifier discrimination. The generator was operated at 40 kV and 40 mA. The sample was prepared for analysis by gently grinding the product obtained in an agate mortar and then depositing on a greased low background sample holder. Diffraction data were collected at room temperature in the range of 6-50° (2θ), in θ:2θ mode and step-scan with ∆2θ=0.03°, for 15 seconds per step.
Structure solution and Rietveld refinement: ZIF-7-III crystal structure was used as a starting model for the isomorphous ZIF-9-III structural refinement. The background and the peak shape were modeled with a 7-term shifted Chebyshev and a pseudo-Voigt function respectively.
Spherical harmonics were used for preferred orientation correction. Benzimidazolate ligands 34 / 36 were treated as rigid bodies to reduce the number of variables. Single isotropic thermal factors were set for the ligands and cobalt element respectively.