Stromaphane: A New Type of Two-dimensional Covalent Organic Framework with Subnanometric Slit-Shaped Pores

Stromaphane is a new type of two-dimensional covalent organic framework (2D-COF), constituting of multiple layers of ladderphanes where two adjacent ladderphane motifs share a common polymeric backbone. The linkers for each ladderphane moiety are perpendicular to the polymeric backbones and staggered with the linkers in the immediate neighbouring ladderphane layers. The framework of the stromaphane contains numerous slit-shaped pores with a width in subnanometres and length in nanometers. This 2D-COF is synthesized by Grubbs G-I catalyst-mediated stereospecic ring opening metathesis polymerisation of a biscyclopropene derivative having a benzene triad linker. The overall process can be considered as a self-intercalation polymerisation of a biscyclopropene. The structure of the 2D-COF has been proved by XRD measurements, DFT simulation, and STM images. This 2D-COF is composed of substituted poly(methylene-E-vinylene) (PMEV) backbones and benzene triad linkers to form a two-dimensional framework with multiply symmetrically distributed subnanometric slit-shaped pores.

staggered with the linkers in the immediate neighbouring ladderphane layers (Figure 1a). Structurally, the framework of stromaphane constitutes of slit-shaped pores with a width in subnanometres and tunable length in nanometres.
Transition metal-catalysed polymerisations have enable diverse chemo-and/or stereo-selective syntheses of one-dimensional polymers. 25 We have extensively applied ruthenium-26 or molybdenum-based catalysts 27 for the stereoselective ring opening metathesis polymerisation (ROMP) 28 of N-arylpyrrolidenefused norbornene or cyclobutene derivatives (NB or CB) leading to the corresponding polymers, PNB and PCB, respectively. Double bonds in PNB are all in trans con guration, when the Grubbs G-I catalyst 26 is used, [28][29][30] and in cis form in both PNB 31 and PCB 32 upon treatment with a Schrock-Hoveyda catalyst SHMo (Figure 1b). 27 The presence of the N-aryl pyrrolidine moiety in NB and CB, is vital for the stereoselective formation of the corresponding polymers PNB and PCB. The pendants in PNB and PCB thus obtained are all aligned towards a similar direction. [29][30][31][32][33] This kind of conformation offers a unique route to synthesise both symmetric and unsymmetrical double stranded ladderphanes (Figure 1a). [21][22][23][24] ROMPs of cyclopropene derivatives have been brie y explored. [33][34][35][36][37][38] In a manner similar to those described above, G-I-catalysed ROMP of cyclopropene spirally linked with N-aryl azetidine (CP) gives 70% yield of N-arylazetidine appended poly(methylene-vinylene) PMEV stereospeci cally with all double bonds in E-con guration together with 25% yield of the corresponding cyclohexadiene derivative CHD-1 ( Figure 1c). 38 Unlike PNB and PCB mentioned above, all of the trans-double bonds in PMEV cause the Narylazetidine pendants to orient up and down in an alternating manner. It seems likely that, during the course of polymerisation of CP, the pendant of the incoming monomeric unit might orient in the direction opposite to that of the pendant immediately next to the ruthenium carbene intermediates E-Ru-carb-i, where i = 1,2,…,n is the number of repeating units (Figure 1c). During the chain-growth process, Z-Ru-carb-(i+2) may also be formed. Since the ruthenium carbene and the e,z-double bond are in close proximity, ring closure metathesis (RCM) will rapidly occur to give CHD-1 and regenerate E-Ru-carb-i, which will continuously undergo chain growth to yield PMEV stereospeci cally with all double bonds in the Econ guration. 38 This stereospeci c synthesis of PMEV is expected to be applicable for the construction of two-dimensional stromaphane SP-1.
We have previously shown that a 3,3-dialkenylazetidine moiety would adopt syn-syn, anti-syn, or anti-anti ( Figure 2) conformation. 38 Since crystallinity is an important criterion for 2D-COF, the alkenyl moieties in stromaphane should adopt the same conformation throughout the polymeric scaffold. Accordingly, as shown in Figure 2, stromaphane may have three possible structures, SP-0-ss, SP-0-as, and SP-0-aa. The spacing between two adjacent N-aryl-azetidine pendants on the same side of the polymeric backbone is estimated to be around 0.75, 0.60, and 0.45 nm for syn-syn, anti-syn, and anti-anti conformations, respectively.

Results And Discussion
Strategy Intercalation polymerisation has been extensively employed for preparing layered nanocomposites. 19,20 However, to the best of our knowledge, the use of this protocol to synthesise 2D polymers has not been reported. It is envisaged that a pathway similar to that shown in Figure 1c might occur, when biscyclopropene BCP-0 with a linker is employed (Figure 3a). Thus, the ROMP of BCP-0 will initially generate stereospeci cally a linear oligomer/polymer (e.g., Tetra-0) that is structurally similar to PMEV.
As shown in Figure 1c, the pendant of the incoming cyclopropene unit and the pendant directly attached to the ruthenium carbene intermediate as in Ru-carb-1 will be in opposite directions. Cross metathesis between these two neighbouring cyclopropene rings on the same side of the polymeric backbone in Tetra-0 is unlikely. Instead, cross metathesis of one of the cyclopropene groups in BCP-0 from the bulk with two cyclopropene rings on the same side of Tetra-0 would lead to the formation of 2D-0. The continuation of these procedures might eventually produce two-dimensional stromaphane SP-0 ( Figure 3a). The overall process can be considered to be the self-intercalation polymerisation of BCP-0.
2D-COFs need rigid frameworks. 7,8 The incorporation of exible linkers into the framework may cause defects in the crystal structure, even though highly crystalline, 2D-COFs containing somewhat exible motifs have recently been explored.  As shown in Figure 2, irrespective of the conformation for alkenyl motifs in stromaphane, the separation between linkers should be in the subnanometre range. Room for the movement of linkers would therefore be fairly limited. It is envisioned that the presence of slightly exible linker might facilitate the intercalation polymerisation.

Synthesis of BCP-1
Biscyclopropene BCP-1 was obtained in 77% yield via the S N 2 reaction of carboxylate CP-1 38 and a,a'dibromo-p-xylene ( Figure 3b). The benzene triad (three benzene rings) in the linker of BCP-1 has limited exibility to adopt the right conformation for the chain growth process.
Self-intercalation polymerisation of BCP-1 Treatment of BCP-1 with 10 mol % of G-I in DCM at ambient temperature for 2 h afforded the corresponding crude polymer (hereafter abbreviated as CDP) in 95% yield ( Figure 3b). Immediately after workup, CDP was initially sparingly soluble in chloroform, but it gradually formed aggregates that were insoluble in any solvent. Accordingly, immediately after CDP was obtained, characterisation and microscopic examinations were performed Gel permeation chromatography (GPC) analysis of CDP shows a bimodal distribution with M n values around 6.0 and 12.2 kD ( Figure S1).
The 1 H NMR spectrum of CDP ( Figure S2) exhibits two broad peaks centred at d 5.80 and 6.10 ppm that were assigned to the absorptions of ole nic protons attributed to PMEV backbone and cyclohexadiene fragments, respectively. The ratio was about 4.3 : 1 after deconvolution analysis of these peaks. It is worth noting that this ratio may not re ect the actual composition because of the solubility problem.
These results suggest that CDP may constitute two polymeric components with one containing PMEV motifs and the other cyclohexadiene moieties. Another possibility would be a single species containing both PMEV and cyclohexadiene functionalities.
As will be discussed in the next two sections, the exibility of the benzene triad linker is an impediment for the resolution of the powder XRD and STM images. Nevertheless, the long range ordered structural pattern offers a strong evidence for crystallinity feature and two-dimensional scaffold.
XRD pattern and the structure of stromaphane SP-1 Under optical microscopy, a sample of CDP was found to consist of crystalline SP-1 ( Figure S3) and amorphous powder AMP-1 ( Figure S4) which were separated manually. A sample of SP-1 in a 0.3 mm capillary tube was subjected to the XRD experiments XRD measurements were performed using beamlines TPS09A and TLS01C2 with wavelengths of 0.082656 nm and 0.103321 nm, respectively. Figure 4a shows the one-dimensional XRD patterns obtained with 0.103321 nm, and the corresponding two-dimensional image pattern is displayed in Figure S5. The results con rm the high crystallinity for SP-1. Further processes (see Methods and Supplementary Information) resulted in having cell constants: a = 6.0455(3) nm, b = 0.92292(4) nm, c = 1.15838(6) nm, and b = 90.649(3) o with monoclinic P a space group and the agreement factors were R wp = 1.17% and R p = 0.72% ( Figure S5).
As shown in Figure 4b, the two nitrogen atoms on the linker anked by the benzene triad are separated by about 2.3 nm and the distance between two benzene rings in the triads is 0.9 nm. The orientation of benzene rings in benzene triads are not homogeneous. The uxional behaviour of these aromatic motifs re ects the presence of a somewhat disordered crystal structure of SP-1. The centre-to-centre distance between the benzene ring of a triad to the corresponding benzene ring of the adjacent triad is in the range of 0.44-0.56 nm. This spacing represents the span of each monomeric unit and is also the width of the slits in SP-1. In comparison with the cartoons in Figure 2, the two geminal ole nic moieties at the C 3position of each azetidine ring in PMEV skeletons may likely adopt anti-anti conformation. The spacing of 0.44-0.56 nm allows fairly little room for the limited movement of the triad linker. Such a minor structural disorder may not signi cantly affect the long-range order of crystallinity in stromaphane SP-1.
In order to avoid steric congestion, the dihedral angle between the two geminal ole nic groups at the C 3position of azetidine ring ranges from 30 to 42 o . In addition, both the simulated XRD structure and the STM image shown in the next section revealed that the triad linker in a ladderphane stripe and the linkers in the two immediate next two ladderphane stripes are in a staggered relationship.
The length of the slit, that is the distance between two quaternary carbon atoms at both ends of the slit, is about 3.3 nm. This spacing would approximately be the separation between two PMEV skeletons in SP-1. Figure 5 shows the STM images of SP-1 on highly ordered pyrolytic graphite (HOPG). The long-range ordered and striped features shown in Figure 5a con rm the simulation results. Figure 5b further con rms the XRD analysis results that indicate that SP-1 is somewhat rigid and 2D-patch-like polymer. The 0.45 nm separation between adjacent triad linkers further suggests that ole nic moieties in the PMEV skeleton may adopt the E-con guration and anti-anti conformation.

STM Images
The stripe directions are attributed to the PMEV backbones, and the angles of approximately 120°b etween the stripes of different patches (Figure 5a) are consistent with the three-fold symmetry of the underlying HOPG. These results suggest fairly strong interactions between the SP-1 and the substrate. Figures 5c and d reveal that, in between stripes, there are other ne structures that are benzene triad linkers. As mentioned above, these linkers are somewhat uxional, and a deviation from the alignment may occasionally occur. In addition, the bright features that may be attributed to the images of benzene triads, are signi cantly smaller than what was expected for the benzene rings. As shown in Figure 4b, some benzene rings are tilted from the plane of the 2D scaffold in the simulated structure for SP-1. Indeed, about 25% variation has been found in the distances between two benzene rings in adjacent linkers in this calculated structure. Figure 5d is superimposed with a model calculated from the XRD results. The interactions between SP-1 and HOPG may also account for the structural discrepancy. Regardless of these minor de ciencies, the structural features shown in Figures 4 and 5 are consistent with a 2D-sheet-like structure for stromaphane SP-1.
As shown in Figure 5a  How can cyclohexadiene be excluded from SP-1?
In the previous sections, both the STM image ( Figure 5) and XRD simulation results (Figure 4) suggest the homogeneous structural feature for SP-1. It is immensely striking to see the selective formation of SP-1 with all double bonds in the trans con guration and anti-anti conformation. As shown in Figure 1c, the initial product such as Ru-carb-2 in the chain growth process adopts the trans-con guration and anticonformation for the newly formed double bond. When the product is a single stranded polymer like PCP, equilibrium may exist between Ru-carb-2 anti and Ru-carb-2 syn or the like and the latter with the synconformation may be thermodynamically preferred. However, this kind of equilibrium might not be particularly favourable for the growth of more rigid 2D-framework of SP-1. In other words, the initially formed anti conformation would continuously generate the corresponding anti conformation throughout the polymerisaton process.
During the course of polymerisation, the intermediate species having a Z-double bond might be theoretically generated along with those with the E-double bond. However, the homogeneous nature of SP-1 suggests that the Z-double bond might not be incorporated in the polymeric scaffold. If the Z-double bond were formed and the fragment containing such double bond were linked to the 2D-framework (Figure 6a), the ruthenium carbene moiety (2D-Z-Ru-1) would be trapped into the sub-nanometric gap between the two adjacent pendants. It is known that tricyclohexylphosphine ligand remains ligating to ruthenium in the ruthenium-carbene-catalysed metathesis process. 26 The spacing allowed for ruthenium carbene in 2D-Z-Ru-1 might not be su cient to contain this bulky ligand.
In a similar manner, the tricyclohexylphosphine ligand would not t into the aperture between the two adjacent pendants in the possible cyclohexadiene-comprising intermediate 2D-CHD-Ru-1. It seems unlikely that the ruthenium carbene species in intermediates similar to 2D-Z-Ru-2 would continue catalysing the chain growth process because of steric congestions. If the positions of Ru-carbene and the Z-double bond are swapped such that the Ru moiety is at the exo position as in 2D-Z-Ru-exo-1 (Figure 6c), severe strain will exist around the Z-double bond.
When ROMP takes place between 2D-E-Ru-2 and a biscyclopropene (such as BCP-1), the initially formed 2D-Z-Ru-2 with the Z-double bond would have an unreacted cyclopropene group at the end of the linker. This fragment would be movable to generate a space to house the tricyclohexylphosphine ligand as in 2D-Z-Ru-2. After ring closure metathesis, the cyclohexadiene-containing fragment CHD-2 would be liberated from the framework to regenerate 2D-E-Ru-2 that would continue the chain growth process leading to SP-1 (Figure 6b). These hypotheses reasonably explain why the cyclohexadiene moiety was absent in SP-1.

Conclusion
We have demonstrated the design, synthesis, and characterisation of a new type of 2D-COF stromaphane SP-1. The uniqueness of this approach includes the rst use of a transition metal catalyst to synthesize a 2D-COF. Under G-I-catalysed ROMP conditions, the cyclopropene derivative CP enables the stereospeci c synthesis of substituted PMEV, while the pendants are arranged up and down in an alternating manner. 38 Based on this preliminary result, a new building block, namely, bis-cyclopropene BCP-1 with a somewhat exible benzene triad linker connected to azetidine moiety, has been synthesized. This unique monomer has paved the way to the discovery of self-intercalation polymerisation to synthesize a new type of 2D-COF, stromaphane SP-1, by Grubbs G-I-catalysed ROMP of BCP-1. The PMEV skeletons with the benzene triad linkers form the basis of the framework for the fairly compact SP-1 The subnanometric slit-shaped pores of SP-1 are symmetrically and homogeneously arranged throughout the planar scaffold. The pore widths were measured and found to be 0.45 nm, whereas the pore length varied depending on the length of the linker. This circumscribed facet of SP-1 stands out as a unique 2D-COF. Although linkers in SP-1 could be fairly exible, but, as mentioned above, the surroundings offer little space for the uxional movement of the linkers. As such, although local defects are present, the long-range-ordered and homogeneous structural features of SP-1 can be clearly seen by XRD and STM examinations. The structural information obtained from XRD and STM image suggests that SP-1 is homogeneous, and that the cyclohexadiene moiety might not be present in SP-1. The proposed mechanism shown in Figure 6 may rationalize why cyclohexadiene is not incorporated in SP-1. The fate of the cyclohexadiene motif, however, remains to be clari ed.
The modi cation of linkers might enable the manipulation of the slit length, and the thus-obtained stromaphane could serve as a potential membrane having homogeneously distributed subnanometric pores for various applications. If the linker is an optoelectronic active oligoaryl moiety, a similar number of chromophores would be well aligned. The behaviour of well-oriented chromophores in these 2D-COF materials for optoelectronic applications might be different from those of small molecules having the same chromophores, but packed in amorphous manner. The present study has thus laid the foundation for further investigations.
Methods XRD determination and simulation of the structure for SP-1 The powder x-ray diffraction experiments were carried out using beamlines TLS01C2 and TPS09A at National Synchrotron Radiation Research Center (NSRRC) in Taiwan. A crystalline sample of SP-1 was placed in a 0.3 mm capillary and the sample was spinning during the course of measurements. The wavelengths used for data collections were 0.082656 nm and 0.103321 nm at beamlines TPS09A and TLS01C2, respectively Figure 4a shows the 1-D XRD pattern measured at 0.103321 nm, and the corresponding 2-D image pattern is displayed in (Figure S5). The XRD pattern indicates the high crystallinity for SP-1, but attempts to collect data by single crystal diffractometer were unsuccessful due to ring diffraction patterns which implied that the quality of the crystal was not good enough. The diffraction peaks of 1-D XRD pattern was initially deconvoluted by pro le tting and the diffraction angles were indexed by DICVOL04 program. 42 The crystal may adopt monoclinic system with cell constants : a = 6.038 (2)       (a) Experimental XRD pattern. The sharp diffraction peaks indicate the long range ordering and crystallinity of SP-1. The rst and strongest diffraction peak is indexed as (2 0 0) based on the obtained cell constants and space group (see context). (b) The packing diagram of DFT optimized crystal structure viewed along b-axis (left) and along c-axis (right). All H-atoms are skipped for clarity. The uxional behavior of the benzene triad linker indicates the exibility and may present somewhat disordered crystal structure, so that the b-axis (~0.92 nm) and a-axis (~6 mm ) are about double the length of estimated monomer moiety adopted in anti-anti conformation ( ~0.45 nm and ~3 nm shown in STM results). In each benzene triad linker, the distances between two nitrogen atoms and two benzene rings are ~2.3 nm and ~0.9 nm, respectively. The distances of benzene centroids between adjacent triad linkers are about 0.44 nm to 0.56 nm. Such two dimensional long range ordering makes a 2-D ladder like structure with a slit length about 3 nm.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.