Catching the Reversible Formation and Reactivity of Surface Defective Sites in Metal–Organic Frameworks: An Operando Ambient Pressure-NEXAFS Investigation

In this work, we apply for the first time ambient pressure operando soft X-ray absorption spectroscopy (XAS) to investigate the location, structural properties, and reactivity of the defective sites present in the prototypical metal–organic framework HKUST-1. We obtained direct evidence that Cu+ defective sites form upon temperature treatment of the powdered form of HKUST-1 at 160 °C and that they are largely distributed on the material surface. Further, a thorough structural characterization of the Cu+/Cu2+ dimeric complexes arising from the temperature-induced dehydration/decarboxylation of the pristine Cu2+/Cu2+ paddlewheel units is reported. In addition to characterizing the surface defects, we demonstrate that CO2 may be reversibly adsorbed and desorbed from the surface defective Cu+/Cu2+ sites. These findings show that ambient pressure soft-XAS, combined with state-of-the-art theoretical calculations, allowed us to shed light on the mechanism involving the decarboxylation of the paddlewheel units on the surface to yield Cu+/Cu2+ complexes and their reversible restoration upon exposure to gaseous CO2.

. Unit cell of HKUST-1 MOF and zoom on the Cu 2+ /Cu 2+ hydrated paddlewheel unit. The Cu 2+ cations are depicted in orange, while the oxygen and carbon atoms are shown in red and grey respectively.   . Representative crystallographic structures of Cu2O and CuO (panels a and b respectively) where the copper cations and oxygen atoms are depicted in orange and red, respectively. Cu L2,3-edge theoretical spectra (panels c and d respectively) calculated from the crystal structures shown above by including all of the atoms within cut-off radii of 4, 5 and 6 Å from the Cu photoabsorber.   Figure S5. Comparison between the Cu L3-edge experimental (black) and theoretical (yellow) spectra relative to the pristine HKUST-1 MOF collected at RT. Constant energy cuts (dotted grey lines) are drawn in proximity of the experimental maxima of peaks A, B and D. The associated molecular cluster is also shown, where the Cu 2+ cation and the oxygen, carbon and hydrogen atoms are depicted in orange, red, black and white, respectively.   Figure S8. Comparison between the Cu L3-edge AP-NEXAFS experimental spectrum (black) of HKUST-1 collected at 160°C after exposure to CO2 and the theoretical spectrum (yellow) calculated for the dehydrated Cu 2+ dimeric complex. Constant energy cuts (dotted grey lines) are drawn in proximity of the maxima of peaks A and D. The associated molecular cluster is also shown, where the Cu 2+ cation and the oxygen, carbon and hydrogen atoms are evidenced in orange, red, black and white colors, respectively. Figure S9. Representative molecular clusters of the hydrated, dehydrated and dehydrated/decarboxylated HKUST-1 MOF dimeric sites (panels a, b and c, respectively) where the Cu 2+ and Cu + cations are depicted in orange and blue, respectively, while the oxygen, carbon, and hydrogen atoms are depicted in red, gray and white, respectively. The Cu L2,3-edge theoretical spectra associated to the dimeric clusters depicted above and calculated within cut-off radii of 3, 4 and 5 Å from the photoabsorber, are shown in panels d, e and f.

Powder X-ray Diffraction (PXRD) Analysis
Gently ground powder of Basolite C300 was deposited in the 2 mm deep hollow of a zero background plate (a properly  Figure S3.

AP-NEXAFS measurements
In order to perform operando Ambient Pressure Near edge X-ray absorption fine structure (AP-NEXAFS) measurements, a specially designed reaction cell has been developed at the APE-HE beamline at the ELETTRA synchrotron radiation source. 12,13,14 The samples inside the reactor cell can be heated from RT up to 400°C ca. and they can be exposed to a flux of several gases at the pressure of 1 bar. TEY mode is used to record the experimental spectra by measuring the drain current from the sample with a picoammeter. The cap of the cell has a Si3N4 membrane (100 nm of thickness) that is transparent to X-rays allowing at the same time the vacuum of the beamline to be preserved; the membrane is electrically isolated from the body of the cell where the sample is located. The measurements were performed keeping the sample grounded and applying a positive bias voltage of 40 V to the membrane. The HKUST-1 MOF powder was fixed on a titanium sample holder and pressed in a pit located onto the holder. The cell was mounted in the ultra-high vacuum chamber of the APE-HE beamline, coaxially with the X-ray beam. The experiments were performed collecting the Cu L2,3-edge spectra in the energy range 930-948 eV at different temperatures and under different gas fluxes (He, H2 and CO2) at 1 bar. Along the gas line a liquid nitrogen trap is placed which eliminates the impurities of water from the gas fluxes. O K-edge spectra were also collected at 160°C in He before and during the CO2 purge to monitor any oxygen products. The energy calibration was carried out by simultaneously collecting the spectra of the given sample and of a CuO reference. The spectral processing was performed using the THORONDOR software: pre-edge and post-edge spline functions were employed to remove the background and to normalize the spectra. 15 Linear and cubic polynomial functions were used for the pre-edge and for the post-edge extraction, respectively. 16 Ultimately, only for the Cu L3 edge, the main peak has been normalized to one to enhance the spectral variations during the chemical treatment.

Theoretical XAS calculations
The Cu L2,3-edge spectra have been calculated using the Finite Difference Method Near Edge Structure code (FDMNES), 17,18 implementing the recently developed sparse solver method. 19 FDMNES uses the density functional theory (DFT) with a local exchange-correlation potential eventually spin-dependent. The spectra were calculated using the multiple scattering theory (MST) including spin-orbit coupling. 17,18 The muffin-tin (MT) approximation was used for the potential and the MT radii were chosen as to minimize the difference between the potential in the MT spheres and in the interstitial region. The Cu L2,3edge theoretical spectra of CuO and Cu2O were calculated starting from the corresponding crystallographic structures 20,21 within cut-off radii of 4, 5 and 6 Å from the photoabsorber ( Figure S4). The main features of the spectra are well reproduced by using the 5 Å cluster and the more distant atoms have been found to provide an almost negligible contribution.
The theoretical spectra of the hydrated HKUST-1 paddlewheel unit, along with those belonging to the Cu 2+ /Cu 2+ and Cu + /Cu 2+ dimers resulting from dehydration and dehydration/decarboxylation of the pristine unit, were calculated starting from the crystal structure of HKUST-1 hydrated 22 obtained from high-resolution synchrotron powder diffraction, within the MST framework. The dehydrated model was obtaining by removing the apical water molecules, while for the decarboxylated cluster a DFT optimization was carried out. Different theoretical spectra have been calculated with increasing cut-off radii from the Cu absorber atom ( Figure S9) and convergence is achieved at 5 Å.
The Cu 2+ theoretical spectra were aligned to the experimental data on the basis of the energy of the main L3-edge peak (peak A, Figure 1a). Conversely, the Cu + theoretical spectra were aligned to the Cu + peak present in the spectrum of the HKUST-1 sample measured in He (50ml/min, 1bar) at 160°C (peak C, Figure 2a). In order to account for the experimental resolution, a Gaussian broadening of 0.7 eV was applied to all the calculated spectra.