Optimisation of Cu+ impregnation of MOF-74 to improve CO/N2 and CO/CO2 separations

Carbon monoxide (CO) purification from syngas impurities is a highly energy and cost intensive process. Adsorption separation using metal–organic frameworks (MOFs) is being explored as an alternative technology for CO/nitrogen (N2) and CO/carbon dioxide (CO2) separation. Currently, MOFs' uptake and selectivity levels do not justify displacement of the current commercially available technologies. Herein, we have impregnated a leading MOF candidate for CO purification, i.e. M-MOF-74 (M = Co or Ni), with Cu+ sites. Cu+ allows strong π-complexation from the 3d electrons with CO, potentially enhancing the separation performance. We have optimised the Cu loading procedure and confirmed the presence of the Cu+ sites using X-ray absorption fine structure analysis (XAFS). In situ XAFS and diffuse reflectance infrared Fourier Transform spectroscopy analyses have demonstrated Cu+–CO binding. The dynamic breakthrough measurements showed an improvement in CO/N2 and CO/CO2 separations upon Cu impregnation. This is because Cu sites do not block the MOF metal sites but rather increase the number of sites available for interactions with CO, and decrease the surface area/porosity available for adsorption of the lighter component.


CO purification performance comparison
. Summary table of MOFs and their Cu impregnated analogues reported to date for CO adsorption and CO/N 2 and CO/CO 2 separation at varying temperature and 1 bar.

Ni-MOF-74 synthesis
Ni-MOF-74 was synthesized based on a previously reported experimental procedure 7 which was modified to increase product yield. 0.933 g nickel(II) acetate tetrahydrate (Ni(OCOCH 3 )2·4H 2 O), 0.373 g H 2 DOBDC (2,5-dihydroxyterephthalic acid), 25 mL THF and 25 mL H 2 O were combined in a beaker and sonicated until dissolved. The contents were transferred into a Teflon liner and sealed inside a stainless steel autoclave. The autoclave and its contents were heated in a convection oven for 3 days at 110 °C. After cooling, the contents were recovered and the supernatant was decanted and replaced by DMF. The product was soaked in DMF for 2 days, replaced with fresh DMF 4 times during this duration. The same procedure was repeated with MeOH. The final product was collected after decanting the MeOH and drying under N 2 flow. Sample activation for characterisation and testing was performed at 250 °C.

Co-MOF-74 synthesis
Co-MOF-74 was synthesized based on a previously reported experimental procedure 8 and modified for sample activation and scaled down for the reaction vessel used. 1.07 g cobalt(II) nitrate hexahydrate (Co(NO 3 ) 2 ·6H 2 O), 0.357 g H 2 DOBDC (2,5-dihydroxyterephthalic acid), 50 mL DMF, 50 mL EtOH and 50 mL H 2 O were combined in a beaker and sonicated until dissolved. The contents were transferred to a 250 mL Schott bottle and heated in a convection oven for 66 h at 100 °C. After cooling, the contents were recovered and the supernatant was decanted and replaced by DMF. The product was soaked in DMF for 2 days, replaced with fresh DMF 4 times during this duration. The same procedure was repeated with MeOH. The final product was collected after decanting the MeOH and drying under N 2 flow. Sample activation for characterisation and testing was performed at 250 °C.

Sample preparation and activation
The samples, after appropriate dilution in boron nitride, were loaded into an open-ended capillary (ø = 3 mm, Length = 100 mm) and mounted on a catalysis testing station, with the capillary connected to gas lines. During the measurements, the samples were heated from room temperature to 250 °C under He atmosphere, stopping at 80 °C and 160 °C to collect higher quality spectra. After reaching the temperature of 250 °C, the atmosphere was changed to 10% CO in He and measurements were collected continuously until no changes of the Cu-edge XANES spectra could be observed. Samples were then brought down to room temperature and XAFS spectra were collected under a CO and N 2 atmospheres.

Data acquisition
XAFS data were acquired at Cu K edge (8.98 keV) for both samples and Ni K edge (8.33 keV) and Co K edge (7.71 keV), for Cu@Ni-MOF-74 and Cu@Co-MOF-74 respectively. Ni and Co edge measurements were performed in transmission mode using ion chambers, while fluorescence mode, acquired using a 64 elements Ge detector, was required for Cu due to the low Cu amount in the sample. Unfortunately it was not possible to collect Cu-edge XAFS spectra under CO:N 2 atmosphere for the Cu@Co-MOF-74 due to experimental issues, but Co-edge XAFS measurement were obtained. A foil of the same element was collected simultaneously to allow for alignment to uniform energy.

Data analysis
XAFS data processing and analysis was performed using the Athena and Artemis software from the Demeter IFEFIT package. 9 Due to the high signal to noise ratio and short acquisition range, only XANES information could be obtained from the Cu edge data. For Ni and Cu, the resultant EXAFS data have been used to determine changes in structure of the MOF structure. The FEFF6 code was used to construct theoretical EXAFS signals, which included single-scattering contributions from atomic shells through the nearest neighboring atoms in the MOF. The k-range used for the fitting went from 3 to 10.634 Å -1 and the r-range from 1.305 to 3.5 Å. The path degeneracy was kept constant in the fit and the amplitude reduction factor (S 0 ) was fixed at 0.800. In particular, the path taken in exam are the M-O and M-M for all spectra, and M-C for the sample under CO atmosphere at room temperature.   Figure S3. Schematic of breakthrough adsorption column instrument for CO/X separation measurements (X = N 2 , CO 2 ). MFC = Mass-flow controller; BPR = Back-pressure regulator; MS = Mass spectrometer.

Breakthrough column measurement calculations
The adsorbate, CO in the example below, can be substituted with N 2 or CO 2 for its corresponding calculations. The dead volume of the system and total voidage of the adsorbent bed must be accounted for in order to calculate the adsorption capacity. Here (Eq 1), overall material balance of the helium can be expressed as the upstream dead volume, downstream dead volume and the total voidage of the adsorbent bed. (1) Overall material balance on the adsorbate: An approximation of the overall material balance: Combining both equations provides the saturation capacity (SC, total adsorption uptake when = at the detector): * Calculated by using the ratio of μg L -1 concentrations of Cu and Ni measured by inductively coupled plasma mass spectrometry. † Calculated by using the ratio of atomic percentage of Cu and Ni measured by X-ray photoelectron spectroscopy.
± The bulk Cu loading, desolvated adsorbent's general formula and crystallographic density was used to calculate the metal density (mol of coordinatively unsaturated metal sites per gram of adsorbent).      Red: oxygen atoms. Purple: nickel atoms. Green: copper clusters (Not to scale). Figure S10. Cu-edge XANES spectra during the initial activation procedure of 4-Cu@Ni-MOF-74 using He and switching to CO. Indicated by arrows are the features for Cu 2+ (Arrow 1) and