Thermal decarboxylation for the generation of hierarchical porosity in isostructural metal–organic frameworks containing open metal sites

The effect of metal-cluster redox identity on the thermal decarboxylation of a series of isostructural metal–organic frameworks (MOFs) with tetracarboxylate-based ligands and trinuclear μ3-oxo clusters was investigated. The PCN-250 series of MOFs can consist of various metal combinations (Fe3, Fe/Ni, Fe/Mn, Fe/Co, Fe/Zn, Al3, In3, and Sc3). The Fe-based system can undergo a thermally induced reductive decarboxylation, producing a mixed valence cluster with decarboxylated ligand fragments subsequently eliminated to form uniform mesopores. We have extended the analysis to alternative monometallic and bimetallic PCN-250 systems to observe the cluster's effect on the decarboxylation process. Our results suggest that the propensity to undergo decarboxylation is directly related to the cluster redox accessibility, with poorly reducible metals, such as Al, In, and Sc, unable to thermally reduce at the readily accessible temperatures of the Fe-containing system. In contrast, the mixed-metal variants are all reducible. We report improvements in gas adsorption behavior, significantly the uniform increase in the heat of adsorption going from the microporous to hierarchically induced decarboxylated samples. This, along with Fe oxidation state changes from 57Fe Mössbauer spectroscopy, suggests that reduction occurs at the clusters and is essential for mesopore formation. These results provide insight into the thermal behavior of redox-active MOFs and suggest a potential future avenue for generating mesoporosity using controlled cluster redox chemistry.


Synthesis of PCN-250 MOF:
To a 20 mL pyrex vial was added H4ABTC (180 mg) and DMF (10 mL). To a second Pyrex vial was added a metal nitrate salt or preformed cluster (540 mg), DMF (2 mL), and glacial Acetic Acid (6mL). The two vials were sonicated until the contents had dissolved. After dissolution, the contents of the vials were combined, filtered, and immediately placed into a 150 °C oven. The MOFs were collected after 48 hours. The products were washed 3x with fresh DMF and allowed to sit in fresh DMF for 24 hours after washing. The samples were then transferred into a soxhlet extractor for two days with Methanol extraction to yield the pure microporous MOF. The samples were dried in air before analysis. Yield: 53%

Generation of hierarchical porosity through decarboxylation:
Generation of hierarchically porous samples occurred through two methods. In the first method, small quantities of the samples, up to 100 mg, were loaded in a BET tube and were heated on the activation port of Micromeritics ASAP 2420 for 12 hours at 185 °C. In order to generate a significant quantity of mesopores in these structures for a larger scale sample (up to 1 g), continual thermal activation under vacuum at 185 °C was utilized on a Schlenk line for 100 hours.

Analysis by Mossbauer sample preparations:
The Iron-containing samples were analyzed via 57 Fe Mössbauer spectroscopy. The thermally activated samples were activated under vacuum and heat on a schlenk line. To ensure no contamination from air, all samples were sealed on the schlenk line and transferred to a glovebox, and sealed in e-icosane for Mossbauer sample preparation. Samples were then transferred immediately into a liquid nitrogen bath for storage after removal from the inert atmosphere once the E-icosane shielding the sample had properly set.

Analysis by LC-MS sample preparations:
Small samples of the thermally treated MOFs were decomposed via 1 mL ammonium hydroxide solution (250ul ammonium hydroxide in 1ml water) at 85 °C for 12 hours. The orange solutions were then filtered and analyzed by LC-MS after dilution with water. Note: ABTC as a free ligand is highly insoluble in most solvents. In order to maintain solubility, the ligand must be in a basic aqueous solution for analysis.

Analysis by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) sample preparation:
Conc. HNO3 (500 μL) was used to dissolve the sample in a vial at 80 °C overnight. A dilution of this was made by first taking 200 μL of this solution and diluting it to 10ml. Then a second dilution was made, taking 100μL from this solution and diluting to 10ml.

Instrumentation
Powder X-ray diffraction (PXRD) was carried out with a Bruker D8-Focus Bragg-Brentano X-ray Powder Diffractometer equipped with a Cu sealed tube (λ = 1.54178 Å) at 40 kV and 40 mA.
Scanning Electron Microscopy (SEM) measurements were carried out on JEOL JSM-7500F. JEOL JSM-7500F is an ultra-high-resolution field emission scanning electron microscope (FE-SEM) equipped with a high brightness conical FE gun and a low aberration conical objective lens.
Thermal Gravimetric Analysis-Mass Spectrometry (TGA-MS) was performed using a Mettler-Toledo TGA/DSC STARe-1 system which was equipped with a GC100 gas controller. The MS data were collected using a ThermoStar Gas Analyzer. The system was sealed from the outside environment during analysis and collected with the carrier gas Ar. N2, CO2, and CH4 sorption measurements were conducted using a Micromeritics ASAP 2020 and 2420 system. The thermal activation before analysis profiles was 185 °C at 10 hours under vacuum for all samples.
Gas Chromatography-Mass Spectrometry (GC-MS) was performed using a Thermo Scientific DSQ II GCMS. An ESI analysis was used.
Inductively Coupled Plasma Mass Spectrometry (ICP-MS) -Calibration standards were prepared from certified reference standards from RICCA Chemical Company. Samples were further analyzed with a PerkinElmer NexION 300D ICP mass spectrometer. The resulting calibration curves have a minimum R2 = 0.9999. Additionally, to maintain accuracy, quality control samples from certified reference standards and internal standards were utilized. The individual results of the triplicate samples were averaged to determine the metal concentration. 57 Fe Mössbauer measurements: Samples were transferred into a cup designed to fit the Mössbauer Spectrometer under an inert atmosphere and were rapidly frozen in liquid N2 in an anaerobic refrigerated glove box. Mössbauer data was then collected at 5K and 150K on an MS4 WRC or LHe6T spectrometer (SEE Co., Edina, MN). The data was analyzed and simulated using WMOSS software (http://www.wmoss.org/). The instrument was calibrated at room temperature with α-iron foil.
For X-ray absorption spectroscopy (XAS), the sample was packed inside the sample holder under a constant flow of high purity helium. Helium flow was used during thermal studies as well. The XAS studies were conducted at the Advanced Photon Source at Argonne National Laboratory under the direction of Dr. Di-Jia Liu.

Discussion of Heat of Adsorption (HOA) Calculations (Figures S43-S50)
Heat of adsorptions for both methane and carbon dioxide were calculated based upon the Langmuir 8-10 fits of the isotherm data at 195 K, 273 K, and 298 K. Langmuir fits for the three data sets were conducted using the OriginPro 8.5 software by taking the original isotherm data points and utilizing the non-linear curve fit to the single-site Langmuir equation: = ( * ) 1 + * P, the pressure (kPa) was set to be the independent variable in the fit, while Q, the quantity adsorbed (cm 3 /g STP), was set as the dependent variable. The three parameters, qsat, K, and n, were varied as part of the fit.
An example of the resulting parameters for the six isotherms and their corresponding error values are shown in the table below: where values for q, b, and v are entered into B5, B6, and B7 respectively, and pressure ranges are entered into column L). The calculated data points for each gas at the three temperatures were then compared to find pressure values with matching adsorption quantities through the use of Excel's Match function ( [MATCH(N210,R:R,0)] as an example. In this, the match function within Excel is finding the location within the row from the 273 K data (Column R) where the value in the row (the quantity adsorbed) matches the quantity adsorbed value in the given row the 298 K data (N210))). For this data, each isotherm was varied only by 0.1 mbar, providing a high degree of data overlap for matching.
In one column, the Match function was utilized to find the quantity adsorbed data row from the 298 K isotherm that matched a given row for the 195 K isotherm quantity adsorbed. In another column, the Match function was utilized in the same manner to match the 298 K and 273 K isotherms.

7
The IF function looks to see if there is a value in row Y (corresponding to 273 K rows with quantities adsorbed matching a 298 K value) and row Z (corresponding the comparable 195 K data) that are greater than 1, if there are, then it gives the row number that corresponds to that location. If both of those things are not true, then the IFNA function responds with a result of "FALSE." We can use the SMALL function on our IFNA row to find the nth smallest number in the row from this info. We utilize a value range column ( =SMALL(AA:AA,A12)to allow us to iterate the small function to get the next smallest value in each subsequent row.
Once we have the row numbers, which correspond directly to quantity adsorbed values for 298 K, we use an INDEX function to search through the entire file and find the data that is in the given row number. In this INDEX function, the $1:$1048576 tells the function that it is looking through the entire sheet, the AB column is a list of row numbers, so AB2 is our first-row number of interest, 14 refers to column N. Together, these two give us cell N210, the value shown, 2.51, is the value given in cell N210. This is the lowest quantity adsorbed value that matches between the three isotherms.
Additional INDEX functions are then used to find the corresponding pressure values for the given quantity adsorbed.
The pressures are located in Column 12 (L) and row 210 (shown in AB2).
Once all three pressure values and their corresponding quantities adsorbed were collected, they were analyzed using the Clausius-Clapeyron 11-13 equation in the form: The natural log of the pressure values was taken and compared against 1/T (1/298, 1/273, and 1/195). This gave us a series of lines with three data points each. We calculated the slope of each line using Excel's SLOPE function, giving us the value for Q/R. We then multiplied this by R (0.008314 kJ/mol*K to give us the value of the heat of adsorption for each quantity adsorbed. We were then able to graph Heat of adsorption (in kJ/mol) versus the quantity adsorbed (in cm 3 /g STP). As shown in Figures S43-S50.                                               The two peaks at 12.16 sec and 12.50 sec match the mass of the H4ABTC ligand. These two peaks correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 The third peak at 13.06 sec matches the thermally decarboxylated ligand. Figure S60. UV-GC Chromatogram of the decomposed sample, PCN-250 (Fe/Co). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected yielding three peaks and each peak was sampled for its component MS spectrum. The two peaks at 12.17 sec and 12.48 sec match the mass of the H4ABTC ligand. These two peaks correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light with this isomerization changing the properties of the molecule. 7 The third peak at 13.28 sec matches the thermally decarboxylated ligand. Figure S61. UV-GC Chromatogram of the decomposed sample of PCN-250 (Fe/Ni). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected, yielding three peaks with each peak being sampled for their component MS spectrum.
The two peaks at 11.92 sec and 12.39 sec match the mass of the H4ABTC ligand. These two peaks correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 The third peak at 12.87 sec matches the thermally decarboxylated ligand. Figure S62. UV-GC Chromatogram of the decomposed sample of PCN-250 (Fe/Mn). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected yielding three peaks, with each peak being sampled for their component MS spectrum.
The two peaks at 12.13 sec and 12.46 sec match the mass of the H4ABTC ligand. The two peaks present here correspond to the cis and trans form of H4ABTC as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 The third peak at 13.02 sec matches the thermally decarboxylated ligand. Figure S63. UV-GC Chromatogram of the decomposed sample of PCN-250 (Fe/Zn). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected yielding three peaks, with each peak being sampled for their component MS spectrum.
The two peaks at 12.12 sec and 12.53 sec match the mass of the H4ABTC ligand. The two peaks present here correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 The third peak at 13.01 sec matches the thermally decarboxylated ligand. Figure S64. UV-GC Chromatogram of decomposed sample of PCN-250 (Sc3). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected yielding three peaks, with each peak being sampled for their component MS spectrum.
The two peaks at 12.06 sec and 12.46 sec match the mass of the H4ABTC ligand. The two peaks present here correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 A third peak corresponding to the mass of the thermally decarboxylated ligand was not detected in this sample. Figure S65. UV-GC Chromatogram of decomposed sample of PCN-250 (Al3). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected yielding three peaks, with each peak being sampled for their component MS spectrum.
The two peaks at 12.21 sec and 12.48 sec match the mass of the H4ABTC ligand. The two peaks present here correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 A third peak corresponding to the mass of the thermally decarboxylated ligand was not detected in this sample. Figure S66. UV-GC Chromatogram of the decomposed sample of PCN-250 (In3). The MOF was decomposed in pH=12 to yield the free H4ABTC ligands in solution. The UV-GC Chromatagram was collected yielding three peaks, with each peak being sampled for their component MS spectrum. The two peaks at 12.18 sec and 12.54 sec match the mass of the H4ABTC ligand. The two peaks present here correspond to the cis and trans form of H4ABTC, as azobenzenes are known to undergo isomerization in the presence of UV light, with this isomerization changing the properties of the molecule. 7 A third peak corresponding to the mass of the thermally decarboxylated ligand was not detected in this sample.