The charge-assisted hydrogen-bonded organic framework (CAHOF) self-assembled from the conjugated acid of tetrakis(4-aminophenyl)methane and 2,6-naphthalenedisulfonate as a new class of recyclable Brønsted acid catalysts

The acid–base neutralization reaction of commercially available disodium 2,6-naphthalenedisulfonate (NDS, 2 equivalents) and the tetrahydrochloride salt of tetrakis(4-aminophenyl)methane (TAPM, 1 equivalent) in water gave a novel three-dimensional charge-assisted hydrogen-bonded framework (CAHOF, F-1). The framework F-1 was characterized by X-ray diffraction, TGA, elemental analysis, and 1H NMR spectroscopy. The framework was supported by hydrogen bonds between the sulfonate anions and the ammonium cations of NDS and protonated TAPM moieties, respectively. The CAHOF material functioned as a new type of catalytically active Brønsted acid in a series of reactions, including the ring opening of epoxides by water and alcohols. A Diels–Alder reaction between cyclopentadiene and methyl vinyl ketone was also catalyzed by F-1 in heptane. Depending on the polarity of the solvent mixture, the CAHOF F-1 could function as a purely heterogeneous catalyst or partly dissociate, providing some dissolved F-1 as the real catalyst. In all cases, the catalyst could easily be recovered and recycled.

. Crystal data and structure refinement parameters for F-1a and F-1a'.

S6
PXRD patterns. Samples F-1 and F-1b were placed between two polyimide films and mounted on a Bruker AXS D8 Advance Vario X-ray powder diffractometer equipped with primary monochromator (Ge111, Cu Kα1, λ = 1.54056 Å) and LynxEye PSD. Data were collected at room temperature. For F-1, the range was 2-90° 2θ. For F-1b, the range was 2-60° 2θ, and the sequence of short scans (≈20 min per full range) was performed to observe the phase transformations. The step size was 0.01° 2θ for both samples. The powder data were processed with Bruker Topas5 software. Phase composition of F-1 was confirmed by Rietveld method ( Figure S1). R values were: Rwp/Rwp'/Rp/Rp'/RBragg = 8.14/9.65/5.99/ 7.23/5.73%. Cell films. An additional effort was made to limit sample contact with air and water vapor. Film edges were greased with fluorinated grease and vacuumed. Diffraction data were collected in a sequence of short (about 20 minutes) runs. The first run is referenced as F-1b start. After 8 hours the diffraction pattern stopped changing. All the runs collected after that were summed up, S7 and the resulting pattern was referenced as F-1b 8h. We expected that without films and grease the phase transformations will be much faster. We were unable to index the pattern F-1b start; the forms of the diffraction peaks with close 2θ and different halfwidths suggested that the sample was a mixture ( Figure S3). The phase composition of F-1b 8h was confirmed by Rietveld method ( Figure S4). R values were: Rwp/Rwp'/Rp/Rp'/RBragg = 7.82/8.76/5.88/ 6.68/5.15%. Cell parameters of F-1b 8h were determined with Pawley method ( Figure S5): a = 20.8677(13)Å, b = 20.0951(13)Å, c = 22.6324(15)Å, β = 92.432(5)°, V = 9482.1(11) Å 3 .
Significant differences in X-ray diffraction patterns of the two powder samples and relatively high R values of the Rietveld fits suggested that some structural parameters (such as water content) varied with the preparation method. Figure S3. Observed (blue), calculated (red) and difference (grey) profiles for the Rietveld refinement of F-1. Figure S4. Observed (blue), calculated (red) and difference (grey) profiles for the Pawley refinement of F-1. Figure S5. Observed (blue) profile for F-1b start. Figure S6. Observed (blue), calculated (red) and difference (grey) profiles for the Rietveld refinement of F-1b 8h. Figure S7. Observed (blue), calculated (red) and difference (grey) profiles for the Pawley refinement of F-1b 8h. S9 Calculation of the four pKa's of TAPM. To check whether the four amonium groups of the aniline moieties affected the pKa's of each other, pKa's of all four protonated states of tetrakis(4-aminophenyl)methane were calculated. All calculations were carried out in Gaussian16 Revision A.03 S3 at M062X S4 /def2TZVP S5 and APFD S6 /def2TZVP levels of theory, using the general methodology described in ref. S7 . We used scaled SAS model S8 with water as a solvent and parameters from S7 , which were obtained for M062X/6-31+G(d,p) level of theory.

S8
Proton solvation free energies were fitted individually for each level of theory to provide correct pKa of aniline (4.61 S9 ); the obtained values are: −267.086 and −270.097 kcal/mol for M062X and APFD, respectively. Calculated pKa's are given in Table S4 and shown in Figure S8. According to Figure S8, pKa's of all four transitions between the protonated states are quite close (within 1.3 pKa units) at both used levels of theory, and we conclude that aniline fragments affect each other rather weakly. This makes the tetrakis(4-aminophenyl)methane the most condensed host of weakly interacting aniline moieties. S10 Figure S8. The calculated pKa's of different forms of protonated tetrakis(4-aminophenyl)methane.
SEM images were obtained with a scanning electron microscope Hitachi TM-1000 with a detector TM1000 EDS, allowing to run local roentgen-spectral analysis (speeding voltage -15 kV). The sulfur atom uniformity of distribution was assessed and proved for different parts of a single particle with a precision of 0.15-0.20%. Figure S9. The particles of F-1 precipitated from the mixture of TAPM and NDS in water.     Figure S14. 13 C NMR spectrum of F-1 in DMSO-d6.

Opening of styrene oxide (2) with MeOH
To a solution of 2 (1.1 g, 9.16 mmol) in 50 mL of MeOH was added F-1 (0,11 g, 0.12 mmol). As one unit of framework contained 4-ammonium groups, the real amount of the catalytic units corresponded to 5 mol % relative to the substrate. The heterogeneous reaction mixture was stirred for 1 hour at a room temperature and then the catalyst was filtered. The filtrate was evaporated and, according to its NMR spectra, did not contain any unreacted 2. The residue was filtered through a pad of SiO2 (EtOAc/hexane, 1/1) and the filtrate evaporated to give 3 as a single isomer in a yield of 80% (  Figure S15. 1 H NMR (400 MHz) spectra of 3. Figure S16. 13 C NMR spectrum of 3 in CDCl3.

S16
The stability of the catalyst was checked in a series of experiments and the spectrum of the recovered catalyst after 5-th run is shown in Figure S17. The same reaction was conducted in a mixture of CH2Cl2 and MeOH at room temperature. 2 (0.2 mL, 1.83 mmol), MeOH (1.48 mL, 36.6 mmol), CH2Cl2 (10 mL), F-1 (0.023 g) were mixed and stirring (700 rpm) was continued for 24 hours. After 24 hours the mixture was filtered and the filtrate was evaporated. The ratio of 2 to 3 was determined by 1 H NMR spectra. S17 Figure S18. 1 H NMR spectrum (CDCl3) of a mixture of 2 and 3 obtained in an experiment run in a CH2Cl2/MeOH solution promoted by F-1.

Opening of styrene oxide (2) with H2O.
The catalyst F-1 (0.23 g) was added to a solution of styrene oxide (2 mL, 1.83 × 10 −2 mol) in 50 mL of CH2Cl2 with stirring, then 100 mL of twice distilled H2O was added, and the mixture stirred (700-1400 rpm) for 24 h at rt. Then the mixture was filtered to remove the insoluble catalyst (0.14 g) and the layers were separated. The aqueous layer was evaporated, the residue taken up in a portion of CH2Cl2 and filtered to remove the insoluble catalyst (0.09 g).

Opening of trans-cyclohexeneoxide (5) with water.
To a solution of cyclohexene oxide (0.92 g, 9.38 × 10 −3 mol) in 25 mL of CH2Cl2 was added F-1 (0.11 g) and then 50 mL of H2O with stirring (700 rpm). The stirring was continued for another 24 hours at a room temperature and then filtered. The transparent organic and aqueous layers were separated and each was evaporated. The aqueous layer residue was dissolved in CH2Cl2 and the mixture was filtered to remove the remaining catalyst. The evaporation of the filtrate gave cyclohexane-1,2-diol in a yield of 80% (0.87 g, 7.5 × 10 −3 mol).  S14 The CH2Cl2 solution was also evaporated and the spectra revealed absence of any epoxide. The evaporation of the solvent gave additionally another 0.1 g (8.6 × 10 −4 mol, 9% mol) of 6. S20 Figure S21. 13 C NMR (101 MHz) of 6.in CDCl3. 10 −3 mol) in MeOH (10 mL) was added F-1 (0.023 g, 2.4 × 10 −5 mol). The reaction mixture was stirred (700 rpm) for 4 h at room temperature. The insoluble F-1was removed by filtration and the filtrate evaporated. To the residue were added p-dinitrobenzene (as a standard, 0.009 g, 5.3 × 10 −5 mol) and CDCl3 (1 mL). The remaining insoluble in CDCl3 F-1was filtered and the 1 H and 13 C NMR spectra of the residue were run ( Figure S23 and S24). The spectra corresponded to the literature data. S15 According to the data the chemical yield of the final product was >98%.    In order to assess the real ratios of the products and the amounts of the remaining initial reagents the evaporation of the layers and the filtration of F-1 had to be avoided. The amounts of the products and the epoxides were estimated directly in the reaction solutions by 1 H NMR with different standrrds to each layer added. The typical experiment is described in 8.1. section.
Then the reaction mixture was left for a period of time to let the layers part. Then the layers were carefully separated and to the upper D2O layer was added t-BuOH (0.011 g, 1.48 × 10 −4 mol). To the bottom CH2Cl2 layer was added (0.0047 g, 2.8 × 10 −5 mol) of p-dinitrobenzene and a small portion was taken from the solution and diluted with CDCl3. It was the sample used to estimate by 1 H NMR the amount of the product in the layer ( Figure S27). The same amount of pdinitrobenzene was added to the middle layer followed by 1 mL of DMSO-d6. Each solution was analyzed by 1 H NMR. The upper layer contained 67% of the diol, the middle fraction had 15% of the diol ( Figure S28) and the bottom layer contained only 13% of the initial epoxide ( Figure   S29). Total yield of 1,2-propanediol was 82%.       9. Opening of propene oxide with H2O. The catalyst F-1 (0.023 g) was added to a solution of 1,2-propene oxide (0.12 mL, 1.83 × 10 −3 mol) in 5 mL of CH2Cl2 with stirring, then 10 mL of twice distilled H2O was added, and the mixture stirred (700-1400 rpm) for 24 h at rt. Then the mixture was filtered and the layers were separated. The insoluble F-1was removed by filtration and the filtrate evaporated. To the residue were added p-dinitrobenzene (as a standard, 0.009 g, 5.3 × 10 −5 mol) and CDCl3 (1 mL). The remaining insoluble in CDCl3 F-1was filtered and the 1 H and 13 C NMR spectra of the residue were run (Figures S36 and S37). The spectra corresponded to the literature data. S16 According to the data the chemical yield of the final product was 56%.    11. Opening of 1,2-hexene oxide with H2O. The same protocol was used as that of sections 8 and 9 with the exception of 1,2-hexene oxide (0.22 mL, 1.83 × 10 −3 mol) being used instead of the previous epoxides and the reaction was run for 6 days at rt. According to its 1 H NMR spectra  27.71, 22.71, 14.00 ppm. The spectra corresponded to the literature data. S18 Figure S40. 1 H NMR (400 MHz) spectra of 1,2-hexanediol in CDCl3. Figure S41. 13 C NMR (101 MHz) of 1,2-hexanediol in CDCl3.
Then the catalyst was filtered and the filtrate evaporated to remove the initial unreacted reagents and the solvent at a reduced pressure to give the mixture of exo-and endo-products (Scheme 3) in a yield of 52% (0.52 g, 3.8 × 10 −3 mol) as a mixture of endo/exo-isomers in a ratio 78  Figure S42. 1 H NMR (400 MHz) spectra of a mixture of endo and exo isomers of the Diels-Alder reaction.