Molecular engineering of organic semiconductors enables noble metal-comparable SERS enhancement and sensitivity

Nanostructured molecular semiconductor films are promising Surface-Enhanced Raman Spectroscopy (SERS) platforms for both fundamental and technological research. Here, we report that a nanostructured film of the small molecule DFP-4T, consisting of a fully π-conjugated diperfluorophenyl-substituted quaterthiophene structure, demonstrates a very large Raman enhancement factor (>105) and a low limit of detection (10−9 M) for the methylene blue probe molecule. This data is comparable to those reported for the best inorganic semiconductor- and even intrinsic plasmonic metal-based SERS platforms. Photoluminescence spectroscopy and computational analysis suggest that both charge-transfer energy and effective molecular interactions, leading to a small but non-zero oscillator strength in the charge-transfer state between the organic semiconductor film and the analyte molecule, are required to achieve large SERS enhancement factors and high molecular sensitivities in these systems. Our results provide not only a considerable experimental advancement in organic SERS figure-of-merits but also a guidance for the molecular design of more sensitive SERS systems.

This manuscript reports a SERS active substrate based on a small fluorine substituted organic molecule. The authors attribute the enhancement mechanism to a weakly absorbing charge transfer state involving the molecules and surface. The prospect of an all organic SERS substrate is exciting and the authors attempt to provide a detailed explanation for this effect. They perform spectroscopic characterization of these systems and report enhancement factors of ca. 10^3-5. This seems promising but I am not sure how unique this response is to the particular system under study. For example, charge transfer interactions between organic and inorganic materials are common which raises the question if this phenomenon is limited to a particular class of materials. All that seems to be required is charge transfer between two organics which is not uncommon. The paper is largely phenomenological in nature although theoretical studies are presented for the sake of estimating CT energies. I would have liked to have seen a more thorough discussion of the proposed mechanism and the uniqueness of the target materials. Comments: -They mention the tendency of DFP-4T molecules to pi stack on the surface as a source of enhancement for Raman cross-sections. This seems reasonable but I wonder if they can elaborate further on the likelihood of a potential plasmonic-type enhancement factor. For example, the delocalized pi-electron densities were cited as a possible avenue for SERS but nothing was discussed mentioning this in the introduction.
-It would be helpful if the authors included absorption spectra in the main text highlighting the proposed charge transfer state. The only evidence was inferred from PL quenching dynamics which can be ambiguous. The combination of two chromophores resulting in quenching is not unique and it is not clear to me how exactly this is related to the proposed SERS mechanism.
- Figure 4a is confusing in terms of the labels. They show a line denoting the excitation wavelength (400 nm) but the respective x-axis intercept shows ~750 nm. Perhaps the labels should be reconfigured to avoid possible confusion?
-Raman enhancement factors are reported but the absolute cross sections at the excitation energy are not provided. - The authors choose to excite the MB analyte off resonance. Why not try exciting on resonance? Was there too much fluorescence background?
Reviewer #3 (Remarks to the Author): The authors of this article build on previous exploration of the Raman enhancement properties of thin films of organic semiconductors. The field of surface-enhanced Raman spectroscopy (SERS) is over 40 years old, having originally been associated with Ag or Au nanoparticles for plasmon resonances coupled to other resonances in the molecule-metal system. The next advance in the field came with the successful exploration of inorganic semiconductors substrates such as metal oxides, sulfides, selenides, nitrides opening a wealth of interesting novel substrates. What these substrates lacked in enhancement power, they made up in stability, cost, and the ability to control more precisely the various resonances in the molecule-substrate system contributing to the enhancement. Only recently have (some of) the current authors pioneered the extension of SERS substrates to organic semiconductors. This area holds considerable promise to extend the range and variability of the possible substrates and the ability to control even more finely the parameters contributing to the enhancement. The authors make considerable use of the theory (both analytical and computational) to guide their choices of substrate.
In this contribution, the authors examine a series of organic semiconductors related to DFP-4T (a fully π-conjugated diperfluorophenyl-substituted quaterthiophene structure) in a thin film. Previous studies on DFH-4T showed enhancement factors around 103, while the end-capped with π-electrondeficient perfluorophenyl DFP-4T used in this study show an enhancement factor greater than 105. This latter is comparable to many of the early metallic substrate SERS experiments. They attribute the increased enhancement to the (usually weak) charge-transfer transition, which has acquired additional intensity due to vibronic coupling with allowed nearby transitions in the moleculesubstrate system. These considerations are augmented by several cogent DFT calculations which reveal the source of the CT transitions and others that pertain to the rather large enhancement. They further show that tuning the location of the CT transition by varying the number of thiophenes (e.g. DFP-nT, where n is varied from 2-5), they can control the location and strength of the CT transition and strongly affect the Raman enhancement.
The authors go a long way towards developing methods to optimize the properties needed for development of sensors using SERS. They are taking advantage of the unique properties of organic semiconductors for Raman enhancement, and are now among the most advanced research groups in this endeavor. This article should definitely be published.
I only have a few comments for improving the work: In the DFT calculations, it appears that the substrate is modeled by utilization of only a single molecule. However, one of the important features of the substrate is the collective nature of the organic aggregate. Many such systems have exciton bands, which are formed in the excited state and might have interesting effects on the enhancement properties. It would be of considerable interest if these were explored by appropriate DFT calculations. If one can make DFT calculations in Ag aggregates of over 1000 atoms, certainly good calculations can be made with a number of DFP-4T molecules. We know the crystal structure, and this can be built into the calculations. Response: In addition to the geometric differences between different semiconductor systems mentioned on page 17 that may account for some of the differences in EFs not captured by our computational model, we have added the following sentences to page 18 to address the reviewer's concern about the enhancement factors on DFPCO-4T and DFH-4T substrates: "In addition, we compute all enhancement factors on resonance with the CT states; if the CT states in the experimental systems somehow differ in how close to resonance they are with the Raman laser, this will likely cause some deviation in the trends in EFs. We note that our computational model reverses the ordering of the enhancement factors on DFPCO-4T and DFH-4T relative to their experimental values, likely due to some combination of the semiconductor/MB molecular packing parameters and resonance conditions described here." 2 Reviewer-1 Comment-2: XRD data of DFP-4T give a distance of 4.17A which does not correspond to any relevant parameter calculated for the crystal. Assuming this distance was representative of the film (not crystal), how would the computed CT quantities and EFs be modified?
Response: The computational study in this manuscript was performed by electronic structure calculations on semiconductor/MB complexes that include a single molecule of the semiconductor and the analyte. Therefore, the intermolecular packing properties of the semiconductor film itself are not taken into account in our computational model. This is mainly because of the high computational cost of the frequency calculations when a large number of semiconductor molecules are included. However, we agree with the reviewer that the packing of the organic substrate is an interesting aspect that definitely deserves further exploration, and this could be the subject of future studies in the field of organic-SERS. To address the reviewer's concern on this matter, the following sentence has been added to page 17: "Because of the high computational cost of the frequency calculations, computations with more than one substrate molecule are not feasible." 3

Reviewer 2
Reviewer-2 General Comments: This manuscript reports a SERS active substrate based on a small fluorine substituted organic molecule. The authors attribute the enhancement mechanism to a weakly absorbing charge transfer state involving the molecules and surface. The prospect of an all organic SERS substrate is exciting and the authors attempt to provide a detailed explanation for this effect.

Response:
We would like to thank the reviewer for his/her favorable comments on our manuscript.
Reviewer-2 Comment-1: They perform spectroscopic characterization of these systems and report enhancement factors of ca. 10^3-5. This seems promising but I am not sure how unique this response is to the particular system under study. For example, charge transfer interactions between organic and inorganic materials are common which raises the question if this phenomenon is limited to a particular class of materials. All that seems to be required is charge transfer between two organics which is not uncommon. The paper is largely phenomenological in nature although theoretical studies are presented for the sake of estimating CT energies. I would have liked to have seen a more thorough discussion of the proposed mechanism and the uniqueness of the target materials.

Response:
We would like to thank the reviewer for raising these important points. There are numerous reasons behind the fact that why organic semiconductors are very promising building blocks to design Raman Enhancement platforms: (i) The presence of energetically accessible and delocalized π-electrons in these systems enables unique chemical, physical, and optoelectronic properties as compared to molecules having only σ-electrons. Particularly, solid-state films of properly designed π-conjugated molecules are responsive to electromagnetic radiation in the visible/near-IR spectral region and conduct electrical current under an external electric field, which brings very exciting opportunities for commercialization. (ii) Since the solid state structure of π-conjugated molecules relies on relatively weak intermolecular interactions (e.g., Van der Waals, π-π, and C-H π), facile film fabrication to form various micro/nano-  Schatz, A. Facchetti, H. Usta, G. Demirel, Nat. Mater. 2017, 16, 918-924.) Prompted by this breakthrough, in this work, we envisioned to design a fluoroarene-modified oligothiophene, 5,5'''-diperfluorophenyl-2,2':5',2'':5'',2'''-quaterthiophene (DFP-4T) semiconductor, which includes an electron-rich quaterthiophene π-core similar to that of DFH-4T but end-capped with π-electron-deficient perfluorophenyl (-C 6 F 5 ) units. Thus, this molecule employs a quite different design approach as compared with DFH-4T, wherein the σ-insulating perfluoroalkyl substituents are replaced with -C 6 F 5 substituents leading to a fully π-conjugated backbone with small intramolecular torsions. Furthermore, this molecular design enhances solidstate molecular packing and correspondingly favors very efficient charge-transfer/transport in films and facilitate frontier orbital wave function overlap with the analyte molecules to further enhance SERS performance. We kept fluorine substituents on the outer benzene rings to continue to provide excellent volatility for reliable and quantitative PVD film fabrication. As a result, nanostructured DFP-4T films prepared via PVD are found to exhibit unprecedented EF values of >10 5 , indicating that the SERS performance of metal-free organic semiconductor films could approach those of the current plasmonic metal and inorganic semiconducting platforms.

5
At this point, we believe that π-conjugated frameworks could be designed to yield either universal or molecule-specific SERS platforms, which would totally depend on the specific chemical structures of the semiconductor and analyte molecules. This point was also discussed by J. R. Lombardi and it's mentioned as one of the key advantages of organic semiconductor SERS platforms as compared with universal coinage metal surfaces (Nature Materials, 2017, 16, 878-880). However, considering that the organic SERS platform is a Interestingly, the CT excited state energies strongly correlate with the substrate HOMO energies.
Within computational error, these are close to the experimental SERS excitation energy (1.58 eV), given the simplifications of the INDO model, suggesting that resonance CT enhancement of the Raman signal is a likely source of the experimentally observed SERS.
Additionally, the high sensitivity of the CT state to molecular geometry allows this state to contribute strongly to the resonance differential polarizability even though its absorption is weak.
Since the oscillator strength strongly depends on the substrate-analyte wavefunction overlap, these results also suggest that obtaining large enhancements is possible only for substrate morphologies that allow significant π-stacking of the analyte with the substrate molecules. The key role of CT state absorption in the Raman enhancement is in line with our experimental observation that a 3D nanostructured organic semiconductor morphology leads to an effective chemisorption of the analyte molecules with proper molecular orientations to facilitate CT. We also note that the porous/shaped 3D morphology of organic semiconductor films would also provide electromagnetic enhancements (probably by a factor of ∼10) (S. Zhou, G. C. Schatz, Chem. Phys. Lett. 403, 62-67, 2005); however, this cannot be the dominant effect as it would lead to much larger intensities for DFP-4T than for MB, which is not what is observed in Figure   2.
As suggested by the reviewer, we have now extended the related discussion about the enhancement mechanism (pages 12-19) and the potential uniqueness of the target molecules (pages 3, 4 and 19) in the revised manuscript.
Reviewer-2 Comment-2: They mention the tendency of DFP-4T molecules to pi stack on the surface as a source of enhancement for Raman cross-sections. This seems reasonable but I wonder if they can elaborate further on the likelihood of a potential plasmonic-type enhancement factor. For example, the delocalized pi-electron densities were cited as a possible avenue for SERS but nothing was discussed mentioning this in the introduction.

Response:
We would like to thank the reviewer for raising this important point. As mentioned in the manuscript, the molecule employed in this work (DFP-4T) is an intrinsic organic semiconductor with a very low intrinsic carrier density of <10 15 cm -3 and negligible plasmon resonances in the conduction band. Therefore, when pristine DFP-4T films are implemented as SERS platforms, despite their unique morphology defined by a rough and nanostructured surface, the electromagnetic mechanism could not play any role. For the current and previous semiconductors studied by us as SERS platforms, chemical enhancement relying on chargetransfer processes between the analyte and semiconductor molecules remains to be the major SERS mechanism. However, it is known in the literature that localized surface plasmons could be generated in non-metallic structures such as inorganic semiconductors (J. Phys. Chem. Lett., 2014, 5, 976-985), and some of these strategies could also be applied to organic semiconductors.
To this end, chemical doping or electrical charge-injection (under gate-bias for example) could tailor band edges and introduce free charge carriers into organic semiconductor structures, which could lead to the formation of localized surface plasmons. This could be potentially useful to further advance the SERS enhancement factors of organic semiconducting platforms. As suggested by the reviewer, we have now discussed these possibilities and potentials in the discussion part of the revised manuscript to provide some guidelines for future studies on organic-SERS platforms (page 19). Reviewer-2 Comment-6: Raman enhancement factors are reported but the absolute cross sections at the excitation energy are not provided.

Response:
We thank the reviewer for his/her comment. In order to calculate the SERS enhancement factor, it is not necessary to know the absolute cross-section of the analyte. Instead, measuring relative Raman intensities and number of molecules in the laser spot areas is sufficient as this is a typically used method in the SERS literature. However, to address the reviewer's concern regarding the absolute cross section of MB on nanostructured DFP-4T surface, we have performed an additional experiment using Rhodamin 6G (R6G) and MB analyte molecules using the same experimental set-up. R6G was selected as a standard to determine the cross-section of MB, and its Raman peak at 1510 cm -1 was used as a reference with a non-SERS absolute crosssection of 9.6x10 -27 cm 2 /sr (E. C. Le Ru, E. Blackie, M. Meyer, P. G. Etchegoin, J. Phys. Chem. C 2007, 111, 137-13803) at 785 nm (extrapolated from 633 nm, assuming absolute cross-section is proportional to υ υ L ) (M. J. Colles, J. E. Griffiths, J. Chem. Phys. 1972, 56, 3384-3391). The absolute SERS cross-section of MB was then determined as 2.4x10 -24 cm 2 /sr for the 1621cm -1 Raman peak. As suggested by the reviewer, we have now included the absolute cross section calculation details in the revised manuscript (page 21).

Response:
We would like to deeply thank the reviewer for his/her favorable comments and support for publication in Nature Communications.

Response:
We would like to thank the reviewer for his/her careful reading of the manuscript.
The typo in Figure 1e has now been corrected as suggested by the reviewer.