Extended Data Fig. 1: Dynamic phase transformations of the buPDI-1P COF monitored with synchrotron radiation (λ = 0.2116 Å).
The sample was exposed to a stream of either toluene-saturated or dry helium. (a) Dynamic phase transformations measured with a time resolution of 12 s per scan. Bottom: Toluene uptake triggers the stepwise transformation from the cp to the ip and the op phases. Top: Desorption of the toluene in dry helium triggers the transformation back to the ip phase. As shown in the toluene sorption experiments, the cp/ip phase transformation occurs at very low toluene concentration and thus was not observed in the time window of this desorption experiment. The phase transitions are apparent from stepwise shifts of the 110 reflection from 0.82° (cp phase) to 0.77° (ip phase) and 0.65° (op phase), respectively. Since the COF does not expand notably along its b axis, the position of the 020 reflection at 0.55° remains virtually unchanged. (b) Non-negative matrix factorization (NMF) analysis of the phase transformations during toluene adsorption confirms that the data set is well represented by three components (these are purely mathematical components). These three components are very close to the simulated patterns for the cp, ip, and op phases (peak intensities differ for some reflections because solvent molecules located in the pores are not included in the simulations). (c) The relative populations of the NMF components showing the evolution and devolution of phases are in very good agreement with the PXRD data shown in (a). Component 1 (representing the cp phase) converts within 40 s to component 2 (ip phase). This component converts to component 3 (op phase) after 80 s of toluene exposure.
