Anisotropic energy transfer in crystalline chromophore assemblies

An ideal material for photon harvesting must allow control of the exciton diffusion length and directionality. This is necessary in order to guide excitons to a reaction center, where their energy can drive a desired process. To reach this goal both of the following are required; short- and long-range structural order in the material and a detailed understanding of the excitonic transport. Here we present a strategy to realize crystalline chromophore assemblies with bespoke architecture. We demonstrate this approach by assembling anthracene dibenzoic acid chromophore into a highly anisotropic, crystalline structure using a layer-by-layer process. We observe two different types of photoexcited states; one monomer-related, the other excimer-related. By incorporating energy-accepting chromophores in this crystalline assembly at different positions, we demonstrate the highly anisotropic motion of the excimer-related state along the [010] direction of the chromophore assembly. In contrast, this anisotropic effect is inefficient for the monomer-related excited state.

1. Novelty and broader interest: As the authors state, it is true that their MOF-structure stands out from existing MOFs in terms of efficient as well as anisotropic energy transfer. Yet, I do not agree with the authors' statement that "the chromophore assembly presented here is distinctly different than the earlier know organic aggregated structures", and that there is a "new type of chromophore assembly". The chromophores in this MOFs form H-aggregates (as evidenced by the spectral features, see Fig. S9), which have been reported for other systems, also for supramolecular assemblies that form by π-π-interactions and/or hydrogen bonding. For the overall photophysics it does not really matter how the structure was formed. Also there is vast amount on reports on excimer formation in supramolecular structures, especially pyrenes are very prone to that. But I guess this a more general discussion, which is be beyond the scope of this paper.
In terms of broader context, in the Conclusions it is stated that MOFs may provide a route to mesoscale crystalline order, which is difficult to achieve for other systems. This is true, but again there is a vast amount of work on organic single crystals (based on small molecules as well as conjugated oligomers and polymers) which have been grown up to many millimetres in size. The authors should report the size of the crystalline domains in their MOFs, and comment on potential routes to increase domain size towards formation of "crystalline films". I find e.g. the statement "... extreme significance towards further development ..." too weak, this should be extended to further highlight the broader interest that this work clearly shows.
2. Exciton annihilation (my previous point 2): Although the fluences used in this work are clearly below those, for which annihilation is typically observed. However, the system presented here is clearly not a "typical" system, it is rather optimised for efficient (Figs. 2 & 4) anisotropic energy transport. As such I believe that annihilation sets in already for fluences significantly below "typical" values, i.e., I would like to see fluence-dependent data to verify the absence of annihilation in the present system. In fact, the observation of a significantly lower annihilation threshold would make this work even stronger. Finally, the fluence reported in the authors' reply should be reported in the paper (I could not find this number).

1.
SEM images of these samples should be given. The AFM images ( Figure S2) showed the roughness of 5 nm for a thin layer of 35 nm in thickness. The sample surface is not smooth at all after comparing the two numbers. Considering the thickness of some samples to be below 20nm, the roughness of 5nm will be considerable and can greatly affect the photophysical measurement. The authors cannot just "trust" literature precedence, as the scientific conclusion of this paper relies on the correct picture more heavily than those in the previous reports.

2.
The monomer excited state cannot be 100% isotropic in its transport in an anisotropic structure. The authors should be careful to not use the word isotropic. I think the authors just want to compare the monomer vs. excimer state. If the authors focus on this comparison to even reflect it in the title, the whole paper will be easier to understand. I strongly suggest the authors to completely rewrite the introduction to focus on the directional transport of excimer excited state vs. monomer excited state.

3.
The difference between the isotropic vs anisotropic model prediction in Figure 4b (black vs. red lines) are very small, while the experimental error can be larger than the difference. This renders the whole discussion and conclusion of this paper unconvincing. I am against the publication of this paper because of this point.

4.
I do not know what normalization method is used for Figure S13, but it seems to show that energy transfer is not present, since the acceptor emission intensity does not increase at donor absorption region but only increases at acceptor absorption region when the doping level goes up.

5.
In the Monte Carlo simulations ( Figure S19), the migration of monomer excited state seems also to be anisotropic, why is it not observed in the experiment?

6.
A key message the paper should but failed to convey is the reason why excimer excited state transport is more anisotropic than monomer excited state transport. The authors have performed simulations but did not extract the key message from these simulations. The general readers want to know the principles behind it.
Reviewer #3 (Remarks to the Author): Haldar et al. present an interesting study on a SURMOF designed to produce anisotropic energy transfer. The experiments appear to be well performed with a good set of controls. The data is well presented, and the interpretation of the results seems consistent. Some cross-sectional SEM of the structures would have been good to have, but this is a minor point. I've also been through all the comments from the reviewers at Nature Materials and the authors response and I think they have done a good job in answering the technical questions raised.
Where I do have reservations about this work is the context in terms of light harvesting. The authors pitch is that their system is very good at anisotropic energy transport. But as reviewer 2 has pointed out, this system is clearly not the only such system out there, nor is it close to being the best. The authors response to reviewer 2 on this point is clearly weak.  (5), pp 648-654). Many of these systems are easier to process and show longer diffusion lengths. The authors system also involves the motion of the monomer exciton, which is not anisotropic. From a device point of view this is not desirable, as its essentially a loss channel.
So overall, while this is a very nice experimental study, I think it will mostly be of interest to sections of the MOF community and is unlikely to have a large impact on the larger community working on light harvesting for energy and sensing applications.

Point-by-point response Reviewer #1 (Remarks to the Author):
In the revised version of the manuscript by Haldar et al., reporting on specifically designed surface-anchored metal-organic frameworks for anisotropic exciton diffusion, my previous issues raised have largely been addressed. Yet, two points require further clarification: We are happy to learn that our answers to the reviewer's comments were mostly satisfactory.
1. Novelty and broader interest: As the authors state, it is true that their MOF-structure stands out from existing MOFs in terms of efficient as well as anisotropic energy transfer. Yet, I do not agree with the authors' statement that "the chromophore assembly presented here is distinctly different than the earlier know organic aggregated structures", and that there is a "new type of chromophore assembly". The chromophores in this MOFs form H-aggregates (as evidenced by the spectral features, see Fig. S9), which have been reported for other systems, also for supramolecular assemblies that form by π-π-interactions and/or hydrogen bonding. For the overall photophysics it does not really matter how the structure was formed. Also there is vast amount on reports on excimer formation in supramolecular structures, especially pyrenes are very prone to that. But I guess this a more general discussion, which is be beyond the scope of this paper.

Response:
We agree with the reviewer in that numerous H-aggregates or excimers are well studied. It is also correct that in some cases they were aggregated into supramolecular structures. However, we feel that our claim of "new type of chromophore assembly" is justified, for the reason provided below: i) In our case, the chromophores form a crystalline structure with well-defined structural parameters. Furthermore, the arrangement of the chromophores leaves a substantial amount of open space (pores within the MOF), which allows for rotational flexibility in the periodic structure. To the best of our knowledge, H-aggregates with these properties have not been described in any work by others. ii) in the H-aggregate, we observe two different excited states, which exhibit quite different properties. The first, an excimer -state, shows an unusual, highly anisotropic motion, while the monomer-state exhibits transport along other directions, too.
The presence of these photophysical features make the chromophore assembly unique and shows a pronounced difference to other known supramolecular systems of H-aggregates. We are not aware of any other works where such anisotropic exciton transport features in a H-aggregate has been observed and carefully characterized.
In terms of broader context, in the Conclusions it is stated that MOFs may provide a route to mesoscale crystalline order, which is difficult to achieve for other systems. This is true, but again there is a vast amount of work on organic single crystals (based on small molecules as well as conjugated oligomers and polymers) which have been grown up to many millimetres in size. The authors should report the size of the crystalline domains in their MOFs, and comment on potential routes to increase domain size towards formation of "crystalline films". I find e.g. the statement "... extreme significance towards further development ..." too weak, this should be extended to further highlight the broader interest that this work clearly shows.

Response:
It is rightly pointed out by the reviewer that the crystalline domain size of these mesoscale ordered SURMOFs is an important parameter that can affect the exciton transport properties. We believe that the exciton diffusion length (~ 97 nm) achieved in the present work can be improved by increasing the crystalline domain size. The crystalline domains in the presented SURMOF range from 50-60 nm. Further development of fabrication technique, such as heteroepitaxial growth of MOFs on metal-hydroxides reported by Falcaro et al (Nat. Mater, 2017, 16, 342) can improve the crystalline domain size. In our revised manuscript, we have added a sentence on this and highlighted in yellow.
The sentence "extreme significance towards further development" has been removed.

Exciton annihilation (my previous point 2):
Although the fluences used in this work are clearly below those, for which annihilation is typically observed. However, the system presented here is clearly not a "typical" system, it is rather optimised for efficient (Figs. 2 & 4) anisotropic energy transport. As such I believe that annihilation sets in already for fluences significantly below "typical" values, i.e., I would like to see fluence-dependent data to verify the absence of annihilation in the present system. In fact, the observation of a significantly lower annihilation threshold would make this work even stronger. Finally, the fluence reported in the authors' reply should be reported in the paper (I could not find this number).

Response:
We agree with the reviewer's comment. Accordingly, we have carried out a new set of measurements and recorded PL spectra in the fluence range of 8.8-380 nJ/cm 2 . These new data (see below) reveal a linear dependence of PL intensity on fluence, suggesting absence of annihilation in this range. Since all measurements on the anisotropic transport were carried out for a maximum fluence of 160 nJ/cm 2 , we can safely conclude that the measured energy transfer efficiencies are unaffected by annihilation effects. A brief discussion of this point has been added to the supporting information of the revised manuscript. At much higher fluence values (>1000 nJ/cm 2 ), annihilation processes can become significant. However, in our work such high fluence excitation has not been used.
Following the reviewers' suggestion we have added the fluence values in the revised manuscript.
3. A minor point: Please provide the photon energies used for excitation in Figs. S10, S11, and S14.

Response:
We have provided the photon energies in the revised supporting information.
After addressing these issues, this very nice manuscript is to my opinion suited for publication in Nature Communications.
We are grateful to this reviewer for the appreciation of the work. We feel we have adequately accounted for all of his/her criticism.

Reviewer #2 (Remarks to the Author):
This manuscript reports the different migration behaviors of monomer and excimer excited states in the SURMOFs systems. Through time-resolved PL measurements and simulations, the authors concluded that the excited state of the excimer transport is anisotropic, but the monomer excited state transport is less anisotropic. The authors revised their previous submission and responded to comments from previous reviewers, but there are still issues left unaddressed, rendering it unready for publication. 1. SEM images of these samples should be given. The AFM images ( Figure S2) showed the roughness of 5 nm for a thin layer of 35 nm in thickness. The sample surface is not smooth at all after comparing the two numbers. Considering the thickness of some samples to be below 20nm, the roughness of 5nm will be considerable and can greatly affect the photophysical measurement. The authors cannot just "trust" literature precedence, as the scientific conclusion of this paper relies on the correct picture more heavily than those in the previous reports.

Response:
We have followed the suggestion of the referee and included a SEM image in the revised supporting information.
We agree with the reviewer that the surface roughness of 5 nm corresponding to the height of ~ 2 unit cells will have an influence on the transport of the excited states in thinner films. Such roughness could account for some of the quenching of the excimer emission. However, this would only make the conclusion that excimer transport is highly anisotropic even stronger as the limited quenching observed in the multilayered structures could be due to interfacial effects. In the future, efforts will be directed towards optimizing synthesis conditions in order to obtain SURMOFs with lower roughness.
2. The monomer excited state cannot be 100% isotropic in its transport in an anisotropic structure. The authors should be careful to not use the word isotropic. I think the authors just want to compare the monomer vs. excimer state. If the authors focus on this comparison to even reflect it in the title, the whole paper will be easier to understand. I strongly suggest the authors to completely rewrite the introduction to focus on the directional transport of excimer excited state vs. monomer excited state.

Response:
The reviewer is correct. In fact, already in the previous version of the manuscript we have stated that the monomer-related state is not truly isotropic. As suggested by the reviewer, we have reworded the manuscript and have avoided the use of the term "isotropic".