3D electron diffraction analysis of a novel, mechanochemically synthesized supramolecular organic framework based on tetrakis-4-(4-pyridyl)phenylmethane

A new supramolecular organic framework (SOF) based on the tetrahedral rigid molecule tetrakis-4-(4-pyridyl)phenylmethane has been mechanochemically synthesized. The crystal structure of the SOF containing benzyl alcohol has been determined using 3D electron diffraction analysis with a novel electron diffractometer.


S2. Powder X-ray Diffraction (PXRD)
Manually selected points were used to describe the background, single crystal data (3D ED) were used to define the unit cell, and cyclic refinements on the entire dataset were used to generate the profile parameters.The peak profile was modelled as a Pseudo-Voight function, corrected due to axial divergence asymmetry and it is cut outside 20*FMWH range.The refinement converged to Rp = 5.77%, wRp = 7.68% and GOF = 2.87 (Figure S1).The indexed cells present a mC Bravais class.

S3.1. Sample preparation
The preparation of a sample suitable for 3D ED analysis was conducted using directly the mechanochemical reaction crude (see S1.2).The nanocrystalline powder of TPPMBnOH was placed on a microscope glass and a TEM sample grid (Cu square grid, 300 mesh, holey carbon film) was gently pressed on it.

S3.2. Electron Diffractometer
3D electron diffraction analysis and scanning transmission electron microscopy imaging have been carried out on an ELDICO ED-1 electron diffractometer.A photograph of the setup is shown below.
The electron gun is made with a self-ramping up beam system having a thermionic LaB6 source which produces a 160 keV electron beam.The beam is shaped by multiple condenser lenses; however, no lens is located close to the sample chamber and no projector lenses are used to project the diffraction pattern onto the detector (Niebel et al. 2021).The condenser lenses are accompanied by several deflectors and among those is a pair of scanning coils that allow STEM imaging.There is only one aperture for both applicable modes, STEM imaging and diffraction.In STEM mode a focussed beam with diameters from 30-40 nm can be realized.The distance of the last condenser lens to the sample is ca.35 mm, which was designed to leave space for further attachments, but also to have a rather large focal length, which means that even out of the focal plane a remarkable depth and sharpness of the images can be achieved, e.g. when images with a large field of view of a tilted or bent grid are recorded.In diffraction mode (parallel beam) recommended beam diameters are in the range from 300 to 950 nm with beam divergences in the range from 0.15 to 0.05 mrad, respectively.
The goniometer, which can mount a standard TEM grid, is equipped with four linear axes and one rotation axis which is vertical (Garbuglia et al. 2022).The construction of the goniometer allows to bring any position on a mounted grid into the point of intersection between rotation axis and the diffraction mode's electron beam (the usual eucentric point in a standard TEM).The movement of the goniometer (especially the sample rotation) is not restricted by any close-by optical elements.The detector is a hybrid pixel detector (Dectris QUADRO) which allows for continuous rotation experiments by its fast readout and neglectable dead time of 100 ns.Due to the fixed sample to detector distance and to the absence of any projection system compared to a TEM, once the detector distance is calibrated with a standard, no daily calibration is necessary and quite precise undistorted measurements of the reciprocal space geometry can be performed retrieving precise unit cell parameters.
The crystals suitable for the 3D ED analysis are searched through STEM imaging and their crystal quality is preliminarily checked with a single diffraction pattern, placing the beam on the crystal of interest.During the analysis, the diffraction patterns are collected while the crystal is rotating, in accordance with the continuous-rotation data collection protocol.

Figure S2
Photograph of the ELDICO ED-1 electron diffractometer setup.The sample chamber has an octagonal shape with several ports, of which the goniometer connection, a window, and the load lock for vacuum transfer are visible.

S3.3. Data Collection and Reduction
Electron diffraction data were collected on different microcrystals illuminated with a parallel electron beam of 750 nm in diameter.The diffraction patterns were recorded with a continuous rotation data collection protocol (cRED), which consists of a continuous scan of the reciprocal space, in this case, up to a maximum of 109° with an angular integration of 0.50° and one second of exposure time per frame.
The position of the object on the TEM grid was assessed through STEM imaging.The eucentric height and the in-plane stability of the crystal were corrected through specific routines implemented in the software integrated with the instrument.Three different crystals were analyzed in total, and two data collections were performed in different regions of the Crystal 2.
The collected diffraction data of each crystal were separately handled using the PETS2 software (Palatinus et al., 2019).The reflection data were indexed also determining the orientation matrix, subsequently the optimization of the frame orientation was performed and the reflection intensities were integrated.Furthermore, the diffraction data were combined using the PETS2 merging tool, which increased the completeness of the reflection data.The reflection intensities were treated with Jana2020 and the symmetrically equivalent reflections were averaged.The merged reflection dataset was used for the ab-initio structure determination, performed by Standard Direct Methods using the SHELXT software (Sheldrick, 2015).The data were initially refined with a fully kinematical approximation, which consists in neglecting the dynamical scattering phenomena and assuming that Ihkl is directly proportional to |Fhkl| 2 .The least-squares refinement was performed with the software SHELXL-2014 (Sheldrick, 2008)  was assessed from the calculation of the difference electrostatic potential map, with a resolution of 0.1 Å.The solvent molecules resulted disordered, so they were refined as a rigid body with an occupancy of 0.25.The amount of BnOH molecules inferred from the refined occupancy is in agreement with the calculated number of electrons from the solvent mask calculation.The solvent mask was calculated with Olex2 on the crystal structure with unmodelled BnOH molecules (Figure S3).The calculation led to 12 electrons per asymmetric unit (theoretical number: 14.5 electrons).
For the dynamical refinement, the reflections were integrated considering a rotation semiangle (Δ) of 0.25°, since it should be one-half of the angular integration step.The integrated intensities were combined in virtual frames with the PETS2 software.Each virtual frame is composed of the sum of 8 experimental frames (NF), for a covered semi-tilt range of 2° (Δv = NFΔ), and the number of overlapping frames (NO) was imposed to 4. The dynamical refinement was conducted on the crystal 2 and crystal 3 data, considering the unit cell obtained from the crystal 2 b data.The dataset from crystal 1 was omitted due to quality reasons.The refinement was conducted simultaneously on the crystal 2 and 3 reflection files using the Jana2020 software.The dynamical refinement was performed after an initial optimization of the crystal thickness and frame geometry.The crystal structure was then refined together with the crystal thickness of the respective crystals of each data collection.

Table S2
Angular coverage of the data collections performed on the analysed microcrystals.The reported values were obtained after a frame-by-frame fine tuning of the geometrical parameters, performed with the PETS2 software.The lowest curve corresponds to the average RC in the range 0.2 to 0.3 Å -1 , the next curves are obtained after a step of 0.1 Å -1 until 0.9 to 1 Å -1 .The red curve is the calculated RC profile while the green line represents the experimental profile.

Table S3
Unit cell parameters obtained from the indexed data of the analysed crystals.

S4. Thermogravimetric Analysis (TGA)
The thermogravimetric analysis of TPPMBnOH was conducted on a Perkin Elmer Instrument, model TGA 8000.The desorption process of the guest molecules was followed in the temperature range 30-350°C with a heating rate of 10°C min -1 under nitrogen flux.The solvate stoichiometry (SS TGA ) was calculated from the weight loss, to define the TPPM/BnOH ratio at atmospheric conditions (Figure S10).The SS TGA of the TPPM .BnOH phase, at room conditions, resulted close to 1.

Figure S12
. Equation adopted for the calculation of the solvate stoichiometry(SS TGA ) from TGA data.

S5. NMR Characterization
The 1 H NMR spectra were collected in CD2Cl2 with few drops of methanol-d4, to obtain a clear solution of TPPM, on a Bruker AC400 Avance.

Figure S1
Figure S1 Profile fit from LeBail refinement on TPPMBnOH.The shown range is limited to 2θ values of 5-50 for clarity, whereas the refinement was carried out in the range 5-75°.The refinement converged to Rp = 5.77%, wRp = 7.68% and GOF = 2.87.

Figure S3
Figure S3 Electron diffraction patterns from the data collection on crystal 3. The figure shows two distinct diffraction patterns at the start (left) and end (right) of the data collection.The blue circle highlights the resolution of 1Å.

Figure
Figure S4 (a): Plot of the shift of the pattern centre.From the centre shift it is possible to distinguish the four datasets used in the data reduction process.(b): Plots of the geometrical corrections applied to each diffraction pattern.ω: angle comprised between the tilt axis and the positive horizontal axis of the image.: tilt angle.β: second tilt angle.(c): Plot of rocking curve (RC) profiles of the merged datasets.

Figure S5
Figure S5 Representation of the solvent mask as a green surface along the crystallographic b-axis direction.

Figure S6
Figure S6 The TPPM0.5BnOH structural model oriented along the crystallographic b-axis.The oxygen atoms are represented in red, carbon atoms in grey, nitrogen atoms in blue and hydrogen atoms in white.The disordered solvent molecules are shown in space filling mode.

Figure S7
Figure S7 Electrostatic potential map of the TPPM structural model, visualized as isosurface within the unit cell borders.The potential map is calculated from the dynamically refined structure by Fourier map calculations.

Figure S8
Figure S8 View of the main C-HN interactions in TPPM0.5BnOH.The N atoms involved in the interaction are represented as blue (N1A) and turquoise (N1B) spheres.The solvent molecules are

Figure S10
Figure S10 View of the TPPM molecule in a ball a stick representation, in which the radius of the sphere is directly proportional to Uiso values.(a) Kinematical refinement; (b) Dynamical refinement.

Figure S11
Figure S11 Thermogravimetric path recorded on the TPPMBnOH crystal phase.The blue line corresponds to the weigh percentage on temperature, while the red dashed line is the first derivative of the weight percentage.

Figure S14 .
Figure S14 .Equation adopted for the calculation of the solvate stoichiometry (SS NMR ) from 1 H NMR data.IBnOH : integral associated to a diagnostic peak of the BnOH molecule; ITPPM : integral associated to a TPPM diagnostic peak; nBnOH : number of hydrogen atoms related to the BnOH diagnostic peak; nTPPM : number of hydrogen atoms related to the TPPM diagnostic peak.

Table S1
Unit cell parameters obtained from the LeBail Refinement on the powder X-ray diffraction data (PXRD).

Table S4
Comparison between the unit cell parameters from 3D ED analysis and from the LeBail refinement on PXRD data.

Table S5
Crystal data and structure refinement for the TPPM0.5BnOHphase from the merged datasets at the kinematical theory level.

Table S6
Crystal data and structure refinement for the TPPM0.5BnOHphase from the dataset simultaneously refined at the dynamical theory level.