Chemicals
All materials were purchased from Sigma-Aldrich, Acros or TCI Europe in the common purities purum, puriss or reagent grade. The materials were used as received without further purification and handled in air unless otherwise noted.
PXRD
PXRD measurements were performed using a Bruker D8 Discover with Ni-filtered Cu Kα radiation and a LynxEye position-sensitive detector. For using PXRD patterns as a merit for the COF crystallinity, measurements were performed with identical parameters, including scan speed and intervals. PXRD patterns presented in Figs. 1c, 3b,d–f and 4b,c were measured at a scan speed of 2 s per 0.05 2θ interval. The remaining PXRD patterns presented in Figs. 3c and 4a and the Supplementary Information were measured at a scan speed of 0.5 s per 0.05 2θ interval. In case of normalization of the diffraction pattern, the point with the highest count score at the pattern has been taken as reference. In Fig. 4b, the area of all crystalline reflections was determined by analysing the respective non-normalized, background-corrected PXRD patterns.
In situ small and wide-angle X-ray scattering
In situ measurements were performed using an Anton Paar SAXSpoint 2.0 system equipped with a Primux 100 micro Cu Kα source and a Dectris EIGER R 1M 2D detector. A reaction solution was prepared before measurement and initiated with acetic acid three minutes before the start of the in situ measurement. This time was needed for transferring the reactive solution into a quartz capillary of 2 mm in diameter and subsequently to the measurement chamber under vacuum. The quartz capillary was kept at 293.2 K during the whole experiment. The capillary was positioned at a sample–detector distance of 180 mm. The data were recorded with intervals of 10 min or 1 h, as indicated in the respective figure captions.
Gas adsorption analysis
Nitrogen sorption isotherms were recorded with Quantachrome Autosorb-1 and Autosorb iQ instruments at 77 K. The samples were outgassed for 24 h at 120 °C under high vacuum before the measurements.
Scanning electron microscopy
Scanning electron microscopy images were recorded with a FEI Helios NanoLab G3 UC scanning electron microscope equipped with a field emission gun operated at 3–5 kV.
Ultraviolet–visible absorption spectroscopy
Ultraviolet–visible spectra were recorded using a PerkinElmer LAMBDA 1050 spectrometer equipped with a 150-mm integrating sphere, photomultiplier tube and InGaAs detectors. Diffuse reflectance spectra were collected with a Praying Mantis (Harrick) accessory and were referenced to barium sulfate powder as white standard. The specular reflection of the sample surface was removed from the signal using apertures that allow only light scattered at angles >20° to pass.
Fourier-transform infrared spectroscopy
The Fourier-transform infrared spectra were recorded on a Bruker VERTEX 70 FT-IR instrument using a liquid-nitrogen-cooled MCT detector and a germanium ATR crystal. The infrared data are background-corrected and reported in frequency of adsorption (cm−1).
COF synthesis procedure in bulk
TA-TAPB COF
Standard protocol with HOAc as catalyst (25 °C/120 °C). The TA-TAPB COF was synthesized following a previous report41. In a 6-ml culture tube, TA (6.05 mg, 0.045 mmol, 1.5 equiv.) and TAPB (10.5 mg, 0.03 mmol, 1 equiv.) were suspended in 500 µl of a 9:1 (v/v) mixture of 1,4-dioxane/mesitylene. Unless otherwise stated, aqueous acetic acid in a specific concentration (6 M, 50 µl) was added, the tube was sealed and the reaction mixture was sonicated to ensure sufficient mixing of the components. Afterwards, the reaction mixture was either heated to 120 °C for 72 h or kept at room temperature for 72 h. After the given reaction time, a yellow precipitate was isolated by filtration under vacuum. Maintaining the material in a slightly wet state24, the sample was extracted using CO2 under supercritical conditions. See also our general comment about drying of the TA-TAPB COF in the following.
Protocol with HOAc/NaCl as catalyst mixture (25 °C). The TA-TAPB COF was synthesized following a previous report41. In a 6-ml culture tube, TA (6.05 mg, 0.045 mmol, 1.5 equiv.) and TAPB (10.5 mg, 0.03 mmol, 1 equiv.) were suspended in 500 µl of a 9:1 (v/v) mixture of 1,4-dioxane/mesitylene. Unless otherwise stated, aqueous acetic acid and NaCl in a specific mixture (3 M HOAc, 1.5 M NaCl, 50 µl) were added, the tube was sealed and the reaction mixture was sonicated to ensure sufficient mixing of the components. Afterwards, the reaction mixture was kept at room temperature for 72 h. Subsequently, the samples were washed with ethanol thoroughly to remove remaining NaCl. Maintaining the material in a slightly wet state24, the sample was extracted using CO2 under supercritical conditions. See also our general comment about drying of the TA-TAPB COF in the following.
Gram-scale protocol with HOAc/NaCl as catalyst mixture (25 °C). The TA-TAPB COF was synthesized on the gram scale according to the lab-scale procedure described above. In a 200-ml Schott Duran glass bottle, TA (1.21 g, 9 mmol, 1.5 equiv.) and TAPB (2.1 g, 6 mmol, 1 equiv.) were suspended in 100 ml of a 9:1 (v/v) mixture of 1,4-dioxane/mesitylene. Aqueous acetic acid and NaCl in a specific mixture (3 M HOAc, 1.5 M NaCl, 10 ml) were added, the glass bottle was sealed and the reaction mixture was sonicated to ensure sufficient mixing of the components. Afterwards, the reaction mixture was kept at room temperature for 72 h. Subsequently, the samples were washed with ethanol thoroughly to remove remaining NaCl and dried under reduced pressure. See also our general comment about drying of the TA-TAPB COF in the following. A yellow powder was obtained (2.49 g; 86%).
General comment on drying of TA-TAPB COF. According to our previous report41, the workup procedure employed can have a big impact on the quality of the obtained TA-TAPB COF. Although we observed in this work that it is possible to obtain the same quality of material by either drying under high vacuum or sCO2 extraction, we recommend sCO2, as it is independent of the human factor and highly robust.
WTA COF
The WTA COF was synthesized following a previous report42. Under argon, N,N,N′,N′-tetrakis(4-aminophenyl)-1,4-phenylenediamine (W, 6.00 mg, 0.013 mmol, 1 equiv.) and TA (3.40 mg, 0.025 mmol, 2 equiv.) were dissolved in 1 ml of benzyl alcohol and mesitylene (9:1 or 10:10 v/v). The aqueous catalyst mixture (50 µl), including acetic acid, acetic acid/NaCl or acetic acid/Ph4PCl in specific concentrations, was added and the tube was sealed. The specific concentrations are stated in the respective Supplementary Figs. After addition of the respective catalyst mixture, the reaction mixtures were kept at room temperature or at 100 °C for 3 days. The resulting red precipitates were collected by filtration, washed with tetrahydrofuran and dried under reduced pressure.
TAPB-DMTA COF
The TAPB-DMTA COF was synthesized according to the synthesis conditions of TA-TAPB COF. In a 6-ml culture tube, 3,6-dimethoxyterephthalaldehyde (DMTA, 4.35 mg, 0.023 mmol, 1.5 equiv.) and TAPB (5.25 mg, 0.015 mmol, 1 equiv.) were suspended in 1 ml of a 9:1 (v/v) mixture of 1,4-dioxane/mesitylene. The aqueous catalyst mixture (50 µl), including either acetic acid or acetic acid/NaCl in specific concentrations, was added and the tube was sealed. The specific concentrations are stated in the respective Supplementary Figs. Afterwards, the reaction mixture was kept at room temperature or 120 °C for 72 h. Subsequently, the samples were washed with ethanol thoroughly to remove remaining NaCl, then with tetrahydrofuran and subsequently dried under reduced pressure. The TAPB-DMTA COF was obtained as yellow powder.
TT-ETTA COF
The TT-ETTA COF was synthesized following a previous report43. Under argon, 1,1,2,2-tetra(4-aminophenyl)ethene (ETTA, 5.85 mg, 0.015 mmol, 1 equiv.) and thieno-[3,2-b]thiophene-2,5-dicarboxaldehyde (TT, 5.9 mg, 0.030 mmol, 2 equiv.) were dissolved in 500 µl of a 9:1 (v/v) mixture of benzyl alcohol and mesitylene. To test the impact of NaCl on the formation of this COF, three different catalyst mixtures were prepared: (1) aqueous acetic acid (6 M, 50 µl); (2) aqueous acetic acid (3 M, 50 µl); and (3) aqueous acetic acid and salt solution (3 M HOAc, 1.5 M NaCl, 50 µl). After addition of the respective catalyst mixture, the reaction mixtures were kept at room temperature for 3 days. The resulting red precipitates were collected by filtration, washed with acetonitrile and dried under reduced pressure. Further, the samples were extracted using CO2 under supercritical conditions to remove solvent residuals.
Operando iSCAT measurements
The synthesis protocol of TA-TAPB COFs on the iSCAT microscope followed the bulk protocols described above and the reaction mixtures were prepared in the same way and at the same concentrations. Unless stated otherwise, for the measurement on the microscope, 100 μl reactant solution (TA-TAPB in 1,4-dioxane/mesitylene) and 10 μl aqueous catalyst mixture (for example, HOAc or HOAc/NaCl) were used. A custom-made reaction cell made from Teflon served as a reactor. In a typical experiment, first the reactant solution was added to the cell. Next, image acquisition is started. After acquiring a sufficient number of images as reference for subsequent background subtraction, the catalyst mixture is added to initiate the reaction.
iSCAT contrast for determining the size of the scatterer
The spatial resolution of an optical technique such as iSCAT is diffraction-limited44. Therefore, in iSCAT, the signal of a subwavelength particle is detected on the camera as a point spread function (PSF). The PSF closely resembles a 2D Gaussian function with a full width at half maximum (FWHM) of roughly half the wavelength of the incident light (that is, FWHMPSF ≫ dscatterer)2. However, in iSCAT, the size of the particle can be retrieved from the contrast (amplitude) of the PSF, which scales with the polarizability and therefore with the volume of the scatterer15. As such, despite being an optical microscope with diffraction-limited resolution, for example, the signal of a 2-nm gold particle can be detected as a PSF in the camera and by analysing its contrast, the size can be determined45. Typically, the contrast–size relationship is calibrated by measuring samples of known sizes15 or through theoretical simulations/calculations46,47.
iSCAT as a method for imaging chemical reactions
Notably, we show that the TA-TAPB COF sample system is not interacting with the illumination light through optical absorption during iSCAT monitoring, accounting for a truly non-invasive measurement (Supplementary Fig. 14). Another important aspect for imaging reactions and the resulting products with iSCAT is that, after a certain film thickness is reached, the camera saturates owing to the high reflection signal. Although this is not relevant for the initial reaction stages, to assess the final growth state, we adjusted the absolute volume of reactant solution (100 µl) to prevent camera saturation. Finally, it has to be noted that the signal contrast in iSCAT relies on refractive index changes rather than chemical specificity, necessitating careful consideration and controls, as reflected throughout the discussion.
Optical setup
The home-built iSCAT microscope was constructed following the instructions in refs. 48,49. The beam path and the different optics used in the setup are described in the following (see also Supplementary Fig. 1). A 785-nm single-mode laser (TOPTICA, iBeam smart 785) acts as the illumination source. The laser is coupled into a single-mode fibre (Thorlabs, custom-made for high-power applications) and is then collimated by a 10× objective (Olympus, Plan N, NA = 0.25). An achromatic plano-convex lens (Thorlabs, AC508-400-AB, f = 40 cm), the focusing lens, focuses the beam into the back-focal plane of a 60× oil-immersion objective (Olympus, PLAPON60X, NA = 1.42). After passing the focusing lens, the beam is directed into the vertically positioned objective by a 45° mirror. The laser beam is collimated by the objective and illuminates the substrate. The backscattered and backreflected light of the illumination beam are then again collected by the oil-immersion objective. After passing the 45° mirror, the reflected and scattered light are sent into the detection arm by a 50/50 beam splitter. In the detection arm, a plano-convex lens (Thorlabs, AC508-1000-A, f = 100 cm), the imaging lens, is installed f = 100 cm after the back-focal plane of the objective and f = 100 cm before a CMOS camera (Photonfocus, MV1-D1024E-160-CL). Therefore, the backscattered light that was collimated after passing the oil-immersion objective is getting focused on the camera detector by the imaging lens. By contrast, the backreflected light that was in focus at the back-focal plane of the oil-immersion objective and was defocusing thereafter gets collimated by the imaging lens and arrives as such at the camera. The signal is emerging as modulation of the reflected light by constructive or destructive interference of the scattered light with the reflected light on the camera. Acquisition software with live processing was custom-made in LabVIEW NXG.
By using a 60× Olympus objective (60× is defined relative to a standard tube lens of f = 18 cm for Olympus objectives) with a f = 100 cm imaging lens, the setup magnification amounts to 333 \(\left(60\times \frac{100\,{\rm{cm}}}{18\,{\rm{cm}}}=333\right)\). One pixel on the detector of the Photonfocus camera has a size of 10.6 µm × 10.6 µm. Considering the magnification of 333, one pixel in a measured image (= one camera detector pixel) corresponds to 31.8 nm \(\left(\frac{10.6\,{\rm{\mu }}{\rm{m}}}{333}=31.8\,{\rm{nm}}\right)\) on the sample/coverglass surface. The standard field of view for images taken this work was 512 × 512 pixels or 16.3 µm × 16.3 µm on the coverglass.
Image acquisition and analysis
In this study, we used the open-source image software Fiji, based on ImageJ, for image-processing purposes. All iSCAT images presented were background-corrected by adopting the temporal median approach. To accomplish this, we calculated the median pixel intensity for each pixel across the first 300 raw images, which were captured at successive time points before the catalyst was added. Unless stated otherwise, images were acquired at a size of 512 × 512 pixels and spatially binned (2 × 2 pixels, resulting in an effective pixel size of 63.6 nm per pixel). Images presented in Figs. 1c and 2b were acquired at a speed of 2.7 ms per frame (exposure time of 1 ms, 370 fps) and five successive frames were temporally averaged, resulting in an effective temporal resolution of 13.5 ms per frame. Experiments presented in Extended Data Fig. 1 were acquired at a speed of 4.7 ms per frame (exposure time of 3 ms, 212 fps). Images presented in Extended Data Figs. 2 and 3 were acquired at a size of 400 × 400 pixels and at a speed of 2.05 ms per frame (exposure time of 1 ms, 488 fps). For images presented in Extended Data Fig. 3c, 100 frames were temporally averaged, resulting in an effective temporal resolution of 205 ms per frame. Images presented in Extended Data Fig. 5b were acquired at a size of 400 × 400 pixels and at a speed of 6.05 ms per frame (exposure time of 5 ms, 165 fps).
