Electronic origins of photocatalytic activity in d0 metal organic frameworks

Metal-organic frameworks (MOFs) containing d0 metals such as NH2-MIL-125(Ti), NH2-UiO-66(Zr) and NH2-UiO-66(Hf) are among the most studied MOFs for photocatalytic applications. Despite structural similarities, we demonstrate that the electronic properties of these MOFs are markedly different. As revealed by quantum chemistry, EPR measurements and transient absorption spectroscopy, the highest occupied and lowest unoccupied orbitals of NH2-MIL-125(Ti) promote a long lived ligand-to-metal charge transfer upon photoexcitation, making this material suitable for photocatalytic applications. In contrast, in case of UiO materials, the d-orbitals of Zr and Hf, are too low in binding energy and thus cannot overlap with the π* orbital of the ligand, making both frontier orbitals localized at the organic linker. This electronic reconfiguration results in short exciton lifetimes and diminishes photocatalytic performance. These results highlight the importance of orbital contributions at the band edges and delineate future directions in the development of photo-active hybrid solids.


Photocatalytic setup
. Schematic representation of the photocatalytic setup employed in this work. 4 N 2 -physisorption experiments were carried out at 77 K in a TriStar II unit gas adsorption analyser (Micromeritics). Prior to the measurements the samples were degassed at 423 K under vacuum for 16 h. The BET areas were calculated using intervals allowing positive BET constants. 1 The total pore volumes were calculated at 0.9 relative pressure.

Powder X-Ray diffraction
Powder X-Ray diffraction patterns were recorded using Bruker-AXS D5005 with Co-Kα radiation.

Thermo Gravimetric Analysis (TGA)
Thermogravimetric analysis was performed by means of Mettler Toledo TGA/SDTA851e, under an air flow of 60 ml min -1 at heating rates of 10 K min -1 up to 1073 K. Figure S5.
The weights of the materials after treatment at 750 o C were converted to 100%, representing the oxidic phase of the used metal. The dashed lines in Figure 1 of the main text represent the weight difference between the oxides and the anhydrous MOFs for the case of stoichiometric crystals. Such differences were calculated assuming that the combustion processes proceed via the reactions (s1)-(s3) and the only remaining solid products are ZrO 2 , HfO 2 and TiO 2 . In this case the nitrogen is accounted for in the form of NO 2 . An example of such calculation is given below: This means that if the mass of the remaining ZrO 2 is assumed to be 100%, the anhydrous stoichiometric NH 2 -UiO-66(Zr) must be positioned at 232% of weight loss scale. The same reasoning was used for calculating the expected weight losses for Ti and Hf.

Structural defects within the NH 2 -UiO-66 frameworks
The UiO-type materials are particularly known for possessing structural defects, as demonstrated by the non-stoichiometric metal to linker ratios observed by several groups experimentally.The missing linkers are often seen being compensated by formates, 3 oxygen, hydroxyl groups, water 2 and/or chloride, 4 however a comprehensive structural information is missing. The influence of these structural defects on gas sorption was documented by Snurr et al. 5 and Zhou et al. 6 In order to consider the influence of the defects on the electronic and photocatalytic properties of NH 2 -UiO-66(X) materials, two synthetic protocols were followed in this work. The first method yields highly defective structures, as documented by Farha and co-workers, whereas the second approach was claimed to produce nearly stoichiometric UiO crystals. These two types of crystals are denoted X d and X i (X = Zr or Hf) and referred to as 'defective' and 'ideal' respectively. Well in line with the previous reports, the lowtemperature protocol results in a defective structure as can be concluded from the BET area exceeding the theoretical prediction for NH 2 -UiO-66(Zr) (see Table S1). The BET area is slightly higher than the predicted value for the case of 8 linkers per node instead of the stoichiometric 12 (1285 vs. 1150 m 2 /g). The structural defects in Zr d are also reflected by the poor crystallinity of the solid ( Figure S3) as well as the small particle size as is evident from the broadening of the Bragg reflections. Scanning electron microscopy confirms these observations as depicted in Figure S4. At the same time the high-temperature synthetic route yields NH 2 -UiO-66(Zr) material, Zr i , possessing high crystallinity, better-distinguishable crystal shape and the surface area closely matching the predicted value for an ideal structure.
The observations for the case of NH 2 -UiO-66(Hf) are similar, yet less prominent. The crystallinity of Hf d is heavily affected by the synthesis procedure. However, the difference in textural properties of Hf d and Hf i is less dramatic than in the case of Zr-based MOFs.
Additional investigation of the defects within the UiO MOFs was carried out using thermogravimetric analysis (TGA). The TGA profiles are depicted in Figure S5. Similarly to the report of Lillerudet al., 4 TGA profiles of the catalysts utilized in this work were normalized with respect to the remaining corresponding oxides. While no clear plateau was observed for UiO-66(Zr) derivatives including NH 2 -UiO-66(Zr) 2 , our experiments allowed for a quantitative TGA analysis. The first weight loss observed up to 110 o C is associated with the solvent removal from the pores of the framework. The second step is the gradual dehydroxylation of the framework, giving rise to Zr 6 O 6 (ATA) 6 followed by the drop at around 400 o C that corresponds to the destruction of the MOF. The decomposition of the ideal sample centre), whereas in the defective MOF the amount of organics is lower. In view of these results, we can conclude that the low temperature synthesis of NH 2 -UiO-66(Hf) does not yield MOF crystals completely free of defects, yet the plateau of the ideal sample is higher, in line with N 2 physisorption results. The presence of structural defects is also typical for Hf-based frameworks of the UiO family. 7,8 The weight of NH 2 -MIL-125(Ti) sample before the full destruction of the framework begins (ca. 350 o C) corresponds well to the theoretical expectations thus suggesting nearly ideal structure of the MOF.

Tauc plot
Diffuse reflectance UV/Vis spectra were collected using a Perkin-Elmer Lambda 900 spectrophotometer equipped with an integrating sphere (''Labsphere") in the 200-800 nm range. BaSO 4 was used as a white standard. The Tauc plot given below was obtained by  Noteworthy, the absorption maxima of the ideal and defective samples are at nearly the same position, although there is a slight blue shift (ca. 7 nm) of the absorption onset in the case of Zr d and Hf d . Another difference when structural defects are present is the reduced intensity of the lowest energy absorption band when compared against the other transition (λ max ca. 250 nm). Since this band is ascribed to the linker excitation, a different linker to metal ratio in the structure could account for the difference in relative intensities. More importantly, the optical spectra of the ideal Zr and Hf MOF pair and those of the defective pair are identical.

EPR light source
EPR spectroscopy. Steady-state EPR measurements were carried out at X-band (9.52 GHz) using a commercial EPR spectrometer Bruker Elexsys E580 equipped with an Oxford Instruments temperature control system (T = 4 -300 K). All spectra were acquired at 40 K. When needed, the samples were exposed to a 500 W mercury lamp equipped with an IR filter (H 2 O, 7 cm optical path) and a UFS6 filter (spectrum given below) for 30 min. After this period the sample tube was placed into liquid nitrogen. The cool-down time to frozen state was ~ 10-15 s. Then it was inserted into EPR cryostat.

External quantum efficiency (EQE)
External quantum efficiencies of catalysts employed in this work were calculated using the following equation: To account for the number of photons required to assemble one hydrogen molecule (2) the reaction rates must be multiplied by 2 and converted to appropriate units to obtain the number of electrons: All the other EQEs were calculated in the same manner.

Transient absorption spectroscopy
Samples were excited using 180 fs pulses at 400 nm for NH 2 -MIL-125(Ti) and 370 nm for NH 2 -UiO-66(Zr) and NH 2 -UiO-66(Hf) generated in a YKGBW oscillator (Light Conversion, Pharos SP) at 1028 nm through nonlinear frequency mixing in an OPA and second harmonics module (Light Conversion, Orpheus). A small fraction of the 1028 nm fundamental beam was split off to generate the broadband probe spectrum in a sapphire (500 -1600 nm) crystal. The probe pulse was delayed relative to the pump using a delay stage with maximum delay of 3 ns. The pump and probe pulses overlap on the sample position under an angle of ~8 degrees, after which the probe light is led to a detector suitable for the probe spectrum selected (Ultrafast Systems, Helios). In order to prevent multiple photons absorption processes, the pump fluence was set sufficiently low, allowing us to study single exciton dynamics. In a typical experiment 2.7 mg of a MOF were dispersed in acetonitrile (700 µL) and sonicated for 30 min. In order to separate large particles ( > 100 nm), the suspension was then centrifuged for 8 min at 10000 rpm. The supernatant was placed in a 2 mm stirred quartz cuvette for the measurements. Transient data were analyzed using a global fitting routine in which the spectral evolution of the time-dependent absorption difference spectra is fitted to a sequential model yielding evolution-associated difference spectra (EADS). 9 In the case of the NH 2 -MIL-125(Ti) and NH 2 -UiO-66(Zr/Hf) 'defective' a three-state kinetic model was used to fit the experimental data. 10 An additional state was used to force the system to go to the ground state.
All time traces were satisfactorily fitted with three time constants of which the last state decays to the ground state being part of the model. For the NH 2 -UiO-66(Zr/Hf) 'ideal' one excited state (which population's decay is defined by exp(-kτ)) that goes to the ground state (described by 1-exp(-kτ)) was sufficient for a good fit.
When only single photon absorption processes occur, a linear correlation between the pump power and the change in absorption should be registered. In order to determine whether multiple photons absorption processes are being observed, a region of the data is selected for every measurement. This region contains the data at the highest intensity; it is averaged for each pump power and then fitted with a linear fit. The resulting relations are referred to as linearity checks and are given below alongside the evolution-associated difference spectra (EADS) and characteristic decays of photoexcited states.