Tuneable near white-emissive two-dimensional covalent organic frameworks

Most two-dimensional (2D) covalent organic frameworks (COFs) are non-fluorescent in the solid state even when they are constructed from emissive building blocks. The fluorescence quenching is usually attributed to non-irradiative rotation-related or π–π stacking-caused thermal energy dissipation process. Currently there is a lack of guiding principle on how to design fluorescent, solid-state material made of COF. Herein, we demonstrate that the eclipsed stacking structure of 2D COFs can be used to turn on, and tune, the solid-state photoluminescence from non-emissive building blocks by the restriction of intramolecular bond rotation via intralayer and interlayer hydrogen bonds among highly organized layers in the eclipse-stacked COFs. Our COFs serve as a platform whereby the size of the conjugated linkers and side-chain functionalities can be varied, rendering the emission colour-tuneable from blue to yellow and even white. This work provides a guide to design new solid-state emitters using COFs.

There is a major concern on the white light emission concept, if the material supposed to be whitelight emission it should be covered their emission spectrum in the entire visible range such as blue, green and red regions more or less in equal proportions (0.33, 0.33, 0.33 in the CIE-1931 chromaticity diagram). However, the obtained CIE coordinates values of reported two best emissive COFs Tf-DHzDAll (0.31, 0.40, 0.39), TFPB-DHzDAll (0.32, 0.46, 0.22) are not that close to the ideal white-light emissive materials. In that case, it is not appropriate to give the statement as "These are very close to the white point of (0.33, 0.33) and can be considered white light emission". Additionally, the visible images of above mentioned COF (TFPB-DHzDAll) display yellow color domination in their PL emission instead of white (in Figure 4a). Further, there is a discrepancy between solid-state PL spectra (or CIE-1931 diagram values, Figure 4b,c) of various COFs and observed PL emission shown in images (irradiated under visible light and 365 nm UV irradiation, Figure 4a), not much distinguish in the emission observed although CIE-1931 and PL spectral displayed clear difference. Therefore, I would recommend the authors to use the laser confocal fluorescence microscope for displaying the true emission from all samples under same excitation.
Few additional comments: (1) FT-IR analysis in the supplementary information is too confusing. It either splits in to 3-4 figures or gives the corresponding chemical structure besides for better understanding for the readers.
(2) Authors should provide more details on the sample preparations, loaded amount and inernel standards (if there is any), etc. for all analytics for example, photoluminescence spectra, lifetime and absolute quantum yield calculations, SEM, TEM, and so on.
(3) Fitting/simulated BET surface area plot of for all the COFs should be given and if possible also mention the value of coefficient of determination.
Reviewer #3 (Remarks to the Author): The MS reports a novel strategy to construct covalent organic frameworks (COFs) that enable photoluminescence (PL) in the solid state. By capitalizing on the flexibility of the linker chromospheres and hydrogen-bond-mediated "freezing" of rotational degrees of freedom, the authors succeeded to color-tune their COFs and even generate white emission from a mixture of appropriately chosen COFs (see Fig. 2). High PL from COFs had been reported previously, as for instance mentioned in refs. 19, 36-38 of the MS, but quenching in COFs containing often-used boronate esters or imine groups in their linkers occurs sometimes due to internal rotational degrees of freedom. Note that ref. 19 solved this problem by appropriate dense design of pi-stacking, while the present MS reports "freezing" these modes that cause internal conversion and non-radiative decay by introducing hydrogen bonds to maintain the planarity of the singly-bonded linkers. In this sense, the work reports a new design strategy, although the individual strategies such as hydrogenbonded freezing has also already reported before (see for instance X. Chen et al, "Locking Covalent Organic Frameworks with Hydrogen Bonds: General and Remarkable Effects on Crystalline Structure, Physical Properties, and Photochemical Activity", J. Am. Chem. Soc. 137, 9, 3241-3247, please insert appropriate citations), and the PL from COFs was also reported before. The versatility of the available linker groups allows the wide range for color tuning, in the end achieving the "holy grail" of white light PL from a mixture of COFs. Therefore, this contribution deserves broad attention in the field, and publication as an article in Nature Communications is recommended after the following minor issues are appropriately addressed.
Being a theoretician, I will restrict my comments to the theory parts only.
1) In the supporting information, when the crystal structure geometry optimization using DFT is reported, it is essential to add the details. What code was used, which functional, what plane wave basis set/pseudopotentials, what was the energy cutoff, what where the convergence thresholds, etc.
2) It would be great if the authors could provide torsional potential energy scans at least in the electronic ground state, such as for instance shown in Chemical Physics Letters 664 (2016) 101-107. Incidentally, this reference should be cited in the paper as it is highly relevant to the presented problem of locking the COF planar structure.

Reviewer 1
We like to thank the referee for the positive endorsement of our work and our responses are included below.
(1) The author said they cannot obtain crystalline COFs constructed with trialdehyde and 2,5-unsubstituted terephthalohydrazide (DHz). But the TFPB-THz and DFDM-THz are obvious difference with DHzDR COFs. There is a references (Chem.Commun., 2014, 50, 12615-12618.) that prepared COFs with Tp and 2,5-unsubstituted terephthalohydrazide (DHz). This COF may own more similar properties with DHzDR COFs. ANS: We have synthesized the TpTh COF using solvothermal method according to Banerjee's work (Response Fig. 1). 1 The TpTh COF is weakly emissive in orange colour with absolute quantum yield of 0.8%. The weak PL is caused by the restriction of intramolecular bond rotation (RIR) via eclipse stacking of the COF. However, due to the absence of 2,5dialkyloxyl group at the hydrazide units, the TpTh COF is incapable of forming intralayer and interlayer hydrogen bonding for RIR, thus it shows weak PL. This example is similar to TFPB-THz and DFDM-THz COFs prepared in our work, which also lack intra-and interlayer hydrogen bonding and display relatively weak PL with absolute quantum yield of 2.4 % and 0.4%, respectively. These results suggest that the intra-and interlayer hydrogen bonding realised by the COF's eclipse stacking is vital to achieve strong PL emission in such materials. (2) The detail information for the structure of prepared COFs need to provide, including the Space group, unit cell dimensions and fractional main atomic coordinates.

Response
ANS: The required data is added in Section Q in supplementary information. Note that for structure determination, the crystal structure was optimized using the Materials Studio Forcite molecular dynamics module with ultra-fine, Universal force fields, Ewald summations condition, and then Pawley refinement was performed using the pseudo-Voigt profile function and Berrar-Baldinozzi function for whole profile fitting and asymmetry correction respectively. To further gain insight of the hydrogen bonding environment in the COFs, more accurate models were relaxed and optimized from the previously refined structures using Density Functional Theory (DFT) with fixed cell parameters but a lower space group (P1).
(See detail in structural modelling method in section A and DFT optimized crystal structure in section N in the supplementary information.)

Reviewer 2
There is a major concern on the white light emission concept, if the material supposed to be  3 Another report showed white light emission of a hydrocarbon nanoring-iodine assembly with CIE coordinate of (0.26, 0.38). 4 To the best of our knowledge, currently there is no strict definition of white light emission in colour science. As shown in Response Fig. 2, the CIE diagram (http://hyperphysics.phy-astr.gsu.edu/hbase/vision/cie.html) reveals different colour regions, which white colour is in the middle cycle. As long as the CIE coordinates fall into this region, we consider it as white emission. We have summarized all the CIE coordinates of COFs into Response Table 1  did not change the results significantly. We used a norm-conserving PBE pseudopotential, with a 60 Ry kinetic energy cutoff, a 1×1×10 Monkhost-Pack k-mesh, and a 10 −6 Ry convergence threshold for self-consistency. For geometry optimization, we used a 10 -4 Ry convergence threshold on total energy and a 10 -3 Ry/au convergence threshold on forces. The vdW-DF2 method was utilized to take the long-range van der Waals interactions into account.
2) It would be great if the authors could provide torsional potential energy scans at least in the electronic ground state, such as for instance shown in Chemical Physics Letters 664 (2016) 101-107. Incidentally, this reference should be cited in the paper as it is highly relevant to the presented problem of locking the COF planar structure.
ANS: We have included the torsional potential energy scan in the electronic ground state in We like to thank the referees for the positive endorsement of our work.

Reviewer 3
One detail remains to be addressed, and that is the name of the exchange functional. DFT contains two functionals, not one, an exchange functional (for instance revised PBEsometimes denoted revPBE or RPB), and a correlation functional (the authors mention they