Metal Node Control of Brønsted Acidity in Heterobimetallic Titanium–Organic Frameworks

Compared to indirect framework modification, synthetic control of cluster composition can be used to gain direct access to catalytic activities exclusive of specific metal combinations. We demonstrate this concept by testing the aminolysis of epoxides with a family of isostructural mesoporous frameworks featuring five combinations of homometallic and heterobimetallic metal-oxo trimers (Fe3, Ti3, TiFe2, TiCo2, and TiNi2). Only TiFe2 nodes display activities comparable to benchmark catalysts based on grafting of strong acids, which here originate from the combination of Lewis Ti4+ and Brønsted Fe3+–OH acid sites. The applicability of MUV-101(Fe) to the synthesis of β-amino alcohols is demonstrated with a scope that also includes the gram scale synthesis of propranolol, a natural β-blocker listed as an essential medicine by the World Health Organization, with excellent yield and selectivity.


S.1.2 Physical and chemical characterization
-X-Ray Diffraction (XRD) patterns were collected in a PANalytical X'Pert PRO diffractometer using copper radiation (Cu Kα = 1.5418 Å) with an X'Celerator detector, operating at 40 mA and 45 kV. Profiles were collected in the 2° < 2θ < 40° range with a step size of 0.017°.
-Scanning Electron Microscopy (SEM) and single point energy-dispersive X-Ray analysis (EDX): particle morphologies, dimensions and mapping were studied with a Hitachi S-4800 scanning electron microscope at an accelerating voltage of 20 kV, over metalized samples with a mixture of gold and palladium during 90 seconds.
-Gas Adsorption measurements were recorded on a Micromeritics 3Flex apparatus at relative pressures up to 1 atm. The sample was degassed overnight at 60 °C and 10 -6 Torr prior to analysis. Surface area, pore size and volume values were calculated from N2 adsorption-desorption isotherms (77 K) Specific surface area was calculated by multi-point Brunauer-Emmett-Teller (BET) method. Total pore volume values were taken at P/P0=0.96. Pore size distributions were analysed by using the solid density functional theory (NLDFT) for the adsorption branch by assuming a cylindrical pore model.
-Fourier-Transform Infrared CO Adsorption experiments were carried out in a Nexus 8700 FTIR spectrophotometer equipped with an infrared cell that allows in situ treatments at controlled temperature and connected to a high vacuum system with gas dosing facility.
-Gas Chromatography with Flame Ionisation Detector (GC-FID) was used to examine the course of the reaction. The equipment was an Agilent 6890N with a column installed of the following characteristics: Agilent 199091J-413 HP-5 with the following dimensions: 30 m x 320 µm x 0.25 µm.
-Inductively Coupled Plasma Mass Spectroscopy (ICP-MS) was performed by digestion in a microwave oven. Analysis was performed in an Agilent 7900 ICP-MS.
-Nuclear Magnetic Resonance (NMR) spectra were recorded on a Bruker Avance III 300 WB spectrometer and were calibrated to the residual solvent peak (CDCl3 at 7.26 ppm for 1 H-NMR and 77.36 ppm for 13 C-NMR). The following abbreviations were used to explain multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet.

S5
The weight loss between 300 and 450 ºC was used to calculate the number of btc linkers per cluster by assuming that: i) any missing btc linker would be compensated by the inclusion of two acetate molecules, ii) acetate molecules would decompose before btc, and iii) after acetate decomposition charge would be compensated by oxide anions. This would agree with the formula: Compared to the theoretical cluster connectivity index of 6, we observe minimum deviations between 5-6 for all sample that rule out significant defectivity changes in these heterometallic frameworks.
Scanning Electron Microscopy (SEM) Figure S3. Scanning Electron Microscopy images of the materials used in this work.

S6
Energy-dispersive X-Ray analysis (EDX) Figure S4. Experimental Ti:M ratio from point and shoot EDX analysis for different crystals. Table S1. Total metallic content determined by ICP-MS. Figure S5. N2 adsorption isotherms of the materials used in this work.

S.3 FTIR-CO ADSORPTION MEASUREMENTS
Approximately 6 mg of the polycrystalline solids were pressed into self-supported wafers and treated under vacuum (10-6 mbar) at 423 K for 5 h. Subsequently, the wafers were cooled down to 118 K under dynamic vacuum followed by CO dosing at increasing pressure. After each CO dosage the FT-IR spectrum was recorded until MOF saturation. Data was baseline corrected using a poly 5 function. Figure S6. FT-IR spectra at 118 K of MUV-101(Fe) after activation at 423 K for 4h (Act.), the introduction of 0.08 mbar of CO (Low P), the introduction of an equilibrium pressure of 288 mbar (High P) and CO desorption through chamber evacuation at 10 -6 mbar (Des.). Act. Des.
High P Low P

S.4.1 Catalytic Procedure
Prior to the reaction, the material is activated overnight in a vacuum oven at 150ºC.
General procedure:

S.4.2 Yield calculation with GC-FID
Aliquots of approximately 50 µL are extracted from the reaction mixture at different times, including the beginning of the experiment, t=0 min, and are diluted using ethyl acetate before centrifugation and injection of the supernatant to the GC-FID. The corresponding yield at each stage was calculated taking into account the initial concentration of cyclohexene oxide with respect to the internal standard.
The chromatograph temperature of injection was 250ºC. The thermal programme of the oven consisted of 2 minutes at 100ºC before a 70ºC/min ramp then holding at 200ºC for a minute, followed by a second ramp of 40ºC/min until 280ºC, temperature at which it stayed for 3.5 more minutes. Using these, isolated peaks of the products were obtained at the following retention times: 4.7 min for cyclohexene oxide, 5.4 min for aniline, 6.7 min for n-Dodecane and 9.8 min for the product 2-(phenylamino)cyclohexan-1-ol (3)

S.5 POISONING OF THE CATALYST
To evaluate the different contributions of each type of acidity towards the catalytic total activity, poisoning tests were carried out. To do so, two different molecules were chosen as probes for each acidic centre. In particular, following the methods described in the literature, 2,6-Lutidine [5] and pyridine [6] were chosen as molecules to specifically interact with Brønsted and Lewis acidic sites respectively thus blocking its catalytic activity.
Experimentally, 100 mg of MUV-101(Fe) were activated under vacuum at 150ºC overnight. Then, to block the Lewis acidic sites, 5 mL of pyridine were added under an Ar atmosphere. The mixture was left undisturbed for 12 hours and the supernatant was removed with a cannula. The material was activated at 110ºC and vacuum overnight. As previously reported, [6] with this activation protocol, pyridine stays coordinated only to Lewis acidic sites.
In order to evaluate the effect of the Brønsted acidic sites only, to the activated MUV-101(Fe), 5 mL of 2,6lutidine were added. After being left still for 12 hours, the supernatant was removed and the solid was activated at 150ºC and vacuum overnight to remove any non-coordinated 2,6-lutidine.
Powder X-Ray Diffractograms and Nitrogen adsorption at 77 K were measured for both solids. The results are shown in Figure S8 and Figure S9, confirming minimum impact to the structure or porosity of the solid after this treatment.

S.7.1 PXRD of MUV-101(M) after catalysis
After 24 hours, the solid was separated through centrifugation from the reaction media, thoroughly washed with ethyl acetate and diethyl ether and let dry in air. The Powder X-ray Diffractograms of the used materials confirm the stability of them in the reaction conditions.

S.7.2 Le Bail Refinement of used MUV-101(Fe)
To carry out the refinement, the Powder X-Ray Diffraction pattern was collected for polycrystalline samples using a 0.5 mm glass capillary mounted and aligned in a PANalytical Empyrean diffractometer (Bragg-Brentano geometry) using copper radiation (Cu Kα λ = 1.5418 A) with an X'Celerator detector, operating at 40 mA and 45 kV. Profiles were collected by using a Soller Slit of 0.02 o and a divergence slit of 1/4 at room temperature in the angular range 2° < 2θ < 80°with a step size of 0.013°.

S.7.4 N2 Adsorption of MUV-101(Fe) after catalysis
Gas adsorption measurements were recorded on a Micromeritics 3Flex apparatus. Prior to analysis, samples were degassed overnight at 150 ºC and 10 -6 Torr. The nitrogen adsorption of the material and the value of BET surface area, were maintained through several cycles, as derives from Figure S17 and Table S4, respectively.

S.7.5 Recyclability tests for MUV-101(Fe)
After each catalytic cycle, the material was washed twice with ethyl acetate and then with diethyl ether to remove any organic substance left in the pore. This washed material was recovered and activated using a similar methodology than that of the fresh one. The evaluation of the recyclability was carried out by comparing the yield obtained in the same reaction conditions after (2% mol MOF and 0.8 mmol reaction scale, neat for 10 hours). As demonstrated in Figure S18, no significant change was found in its catalytic activity, in agreement with the maintenance of the MOF structure previously observed.

ICP
ICP-MS was performed to the supernatant of the reaction after 24 hours. Following the general procedure for the catalysis, 80 µL of cyclohexene oxide (0.80 mmol), 80 µL of aniline (0.85 mmol) and 80 µL of dodecane (0.35 mmol) were added to a vial containing 10 mg of previously activated MUV-101(Fe). The reaction was left stirring at 30ºC under neat conditions for 24 hours. After this time, in order to study the catalyst stability in the reaction media, the supernatant was separated through centrifugation at 8000 rpm during 10 min. The concentration in mg/L of the different metals found using this technique are presented in Table S4.

Hot-Filtration Test
A Hot-Filtration Test (HFT) was carried out to confirm the heterogeneous nature of the reaction with MUV-101(Fe), ruling out the possibility of leached metal ions acting as catalytic centres. In order to conduct the experiment, the reaction was carried out following the general procedure for 2% MOF loading (80 µL of aniline, cyclohexene oxide and n-dodecane at 30ºC). The mixture was stirring for 45 min. After that time, the supernatant of the reaction was separated through centrifugation at 8000 rpm during 10 min. Then, the solution was filtered and placed in a new vial and its evolution followed through GC-FID at different times. As can be seen from Figure S19, the yield of the desired product (3) did not evolve over time in the absence of catalyst.

S.8. SCOPE OF THE REACTION
Following the general procedure for the 5% MOF loading (see section S.4.1), 0.34 mmol, (1 equiv.) of the corresponding amine, 0.32 mmol, (1 equiv.) of the corresponding oxirane and 80 µL of n-Dodecane as internal standard were added to 2 mL vial containing 10 mg of activated MUV-101(Fe) (5% mol MOF). The mixture was stirring during 4h at 30ºC. Subsequently, the crude was dissolved in a minimal amount of DCM and purified by flash column chromatography using hexane and ethyl acetate as eluent.
For GC-FID yields, after 2/4 hours, an aliquot was extracted from the reaction media, diluted with ethyl acetate and centrifuged before being injected in the Gas Chromatograph in order to obtain the yield. Table S6. Different epoxides and amines used to explore the scope of the reaction.