Geometry Change of 1,3-Dicyanobenzene upon Electronic Excitation from a Franck–Condon Fit of Several Fluorescence Emission Spectraâ€
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
The manuscript by Martini, et al uses the Franck-Condon principle to fit fluorescence emission spectra and gain insight to the geometry change of the 1,3-DCB molecules when it is excited to the singlet state from ground state. The results shows good agreement between experiment and theoretical prediction. It is a good fit for the journal, but it needs to be revised for several points:
- In the introduction, the authors are trying to discuss the progress of Dicyanobenzene studies in last four paragraph, but it ends abruptly and audience will be lost. The content needs rearranged or reorganized (maybe should be put somewhere before the authors introducing this paper's work?) so that it can lead to the motivation of this study. The authors may elaborate more on the motivation of their work also.
- Adding the laser excitation wavelength information for fig 3?
- Table S3, fig s2-S6 never appear in main text. Needs some hooks or simple description in main text to refer the audience to the SI.
- Figure 4, the peak intensity fit between experiment and simulation are relatively close in 1-2000 cm-1, but in the range after 2000 cm-1, the intensity in simulation is pretty high, is there some explanation?
Minor points:
- 1st line after 'computational results': ,3-DCB has has C2v-symmetry, double 'has'
- 1st line after Figure 1, 'The transition dipole moment (TDM) of 1,3-DCB is calculated to be oriented along the
a-axis and perpendicular to the c-axis, as shown in Figure 2'. To be more clear, it is 'the TDM of S1 state'
- 3 table in SI has same label: table S5
Author Response
Comment 1: In the introduction, the authors are trying to discuss the progress of Dicyanobenzene studies in last four paragraph, but it ends abruptly and audience will be lost. The content needs rearranged or reorganized (maybe should be put somewhere before the authors introducing this paper's work?) so that it can lead to the motivation of this study. The authors may elaborate more on the motivation of their work also.
Answer 1: We rearranged the introduction according to the suggestion of the reviewer.
Comment 2: Adding the laser excitation wavelength information for fig 3?
Answer 2: done
Comment 3: Table S3, fig s2-S6 never appear in main text. Needs some hooks or simple description in main text to refer the audience to the SI.
Answer 3: done
Comment 4: Figure 4, the peak intensity fit between experiment and simulation are relatively close in 1-2000 cm-1, but in the range after 2000 cm-1, the intensity in simulation is pretty high, is there some explanation?
Answer: The fitting process of FCFit mainly relies on ab initio calculated molecular geometries and their second derivative (Hessian) as well as the experimentally determined line intensities and the distortion of ground state vibrational modes, the latter having a strong influence on the final shape of the fitted spectrum. Additionally, changes in rotational constants upon excitation can be factored in if they are available, which is not the case for this publication.
The final fit is a result of an iterative process with many variables and possible combinations with the goal of matching of the general shape of the fitted spectrum to the experiment while also correctly determining the excited state geometry. Single excited mode intensities can generally be fitted with higher precision than overtones or combination modes and small deviations can occur.
Minor points:
1st line after 'computational results': ,3-DCB has has C2v-symmetry, double 'has'
Answer: Corrected
1st line after Figure 1, 'The transition dipole moment (TDM) of 1,3-DCB is calculated to be oriented along the a-axis and perpendicular to the c-axis, as shown in Figure 2'. To be more clear, it is 'the TDM of S1 state'
Answer: Added missing detail
3 table in SI has same label: table S5
Answer: These are one table spanning over multiple pages. We changed the caption on the following pages to “S5: continued” to clarify this.
Reviewer 2 Report
Comments and Suggestions for Authors
This is a nicely written manuscript from Michael Schmitt et al. that presents the application of Franck-Condon (FC) fits to the characterization of the change of the geometry of 1,3-Dicyanobenzene upon electronic excitation to the lowest excited singlet state. Compared the results of FC fits with the results of ab initio calculations, there was a good agreement between the FC fits and experimental data. This study shows that analytical technique towards spectra could be used in revealing the fingerprints of the system. I think this manuscript could be appealing enough to readers of “Spectrosc. J.” and be published by “Spectrosc. J.” with minor revisions.
1. The authors mentioned that the vibronic intensities in fluorescence emission spectra were fitted with the Franck-Condon (FC) or the Franck-Condon-Herzberg-Teller (FCHT) approximations in different approaches. In this manuscript, the FC analysis was employed. How about the HT effects in the system upon photoexcitation? If the fitting results deviate significantly from the experimental results, is it then necessary to consider the effect of HT?
2. Comparing the simulated spectra with the experimental spectra in Figure 4, some of the vibronic peaks have large differences in intensity, such as Q202 and Q20+Q24. I think it would be better to add some paragraphs describing these differences.
3. Focusing on vibronic couplings in photoexcitation systems, several two-dimensional techniques such as two-dimensional electronic spectroscopy (2D-ES) and two-dimensional electronic-vibrational (2D-EV) spectroscopy were developed and applied. Whether the FC analysis in the manuscript also be used in the analysis of the results of these methods?
4. After Figure 6, the impact of structural differences on the corresponding modes needs to be described in more detail.
Author Response
Comment 1: The authors mentioned that the vibronic intensities in fluorescence emission spectra were fitted with the Franck-Condon (FC) or the Franck-Condon-Herzberg-Teller (FCHT) approximations in different approaches. In this manuscript, the FC analysis was employed. How about the HT effects in the system upon photoexcitation? If the fitting results deviate significantly from the experimental results, is it then necessary to consider the effect of HT?
Answer 1: Yes, indeed, there are systems, in which a perturbing state is near the S1, leading to considerable HT mixing and intensity stealing. In these cases HT corrections have to be made. They comprise the calculation of the numerical derivatives of the transition dipole moment along the normal modes. Once this information is available, it is straightforward to enter the derivatives into the FCFit program. The numerical calculation of the derivatives, however, is very tedious, since it requires 2x(3N-6) full CC2 calculations. We therefore limit this option to cases, in which the perturbation is so large that it causes visible deviations of the intensities. As we didn't use the option in the current contribution, we added the information on the prerequisites for HT corrections to the online supplementary information.
Comment 2: Comparing the simulated spectra with the experimental spectra in Figure 4, some of the vibronic peaks have large differences in intensity, such as Q202 and Q20+Q24. I think it would be better to add some paragraphs describing these differences.
Answer 2: This comment is absolutely true for the simulation. The fit of the S1 geometry shows a very good agreement. It seems that the ab initio geometry is distorted in a manner, which resembles the Q20 mode. We added a respective comment: "It is remarkable how the fit of the geometry change improves the intensity features compared to the simulation, performed at the \textit{ab initio} geometry of the excited state. This is especially true for the overtone of mode Q$_{20}$ and combination bands of mode Q$_{20}$. Obviously, the \textit{ab initio} geometry is distorted considerably along this mode compared to the ground state in contrast to the experimental finding."
Comment 3: Focusing on vibronic couplings in photoexcitation systems, several two-dimensional techniques such as two-dimensional electronic spectroscopy (2D-ES) and two-dimensional electronic-vibrational (2D-EV) spectroscopy were developed and applied. Whether the FC analysis in the manuscript also be used in the analysis of the results of these methods?
Answer 3: In principle yes, but of course the cost function has to be adapted. 2-D techniques yield much more than intensities, so I suppose cross-correlation between the computed und measured spectrum can be used. The changes to be applied to the current program would be massive.
Comment 4: After Figure 6, the impact of structural differences on the corresponding modes needs to be described in more detail.
Answer 4: We added the following: "All ring CC bond lengths increase between 2.6 and 4.0 pm, what can be expected for a $\pi\pi^{\ast}$ excitation. The largest difference between \texit{ab initio} calculated and experimentally determined ring bond lengths is found for the C(4)-C(5) and C(5)-C(6) bonds. Here, the theoretical and measured bond lengths differ by 5 pm, the former being longer than the latter. It is this difference, that can be compensated by a combination of the modes Q$_{20}$ and Q$_{24}$."
Thank you very much for your good points and helpful comments, that helped to improve the manuscript.
Reviewer 3 Report
Comments and Suggestions for Authors
Manuscript number: spectroscj-3261707
The paper titled as “Geometry Change of 1,3-Dicyanobenzene upon Electronic Excitation from a Franck-Condon Fit of Several Fluorescence Emission Spectra” has theoretically introduced the comparison between geometry changes (obtained from the FC fts) and the results of ab initio calculations at the SCS-CC2/cc-pVTZ level. Additional language polish and content revised still need to be emphasized before been published on this journal.
1. Can you explain how the Duschinsky matrix facilitates the understanding of vibrational coupling between the ground and excited states? What insights does it provide regarding the vibrational modes in the S1-state? (page 7, paragraph 1)
2. What significance do the changes in the CN bond length have on the overall electronic and geometric properties of the chromophore? How might these changes affect the molecule's photophysical behavior? (page 11)
3. How do the results from the SCS-CC2/cc-pVTZ level of theory compare with other computational methods (e.g., DFT, MP2)? Are there specific advantages or limitations of the SCS-CC2 method in predicting bond length changes? (page 11)
4. What are the implications of the symmetric increase in aromatic ring CC bond lengths for the electronic and vibrational properties of 1,3-DCB? How might this affect its photochemical behavior?
5. For easier reading and understanding, the introduction and content should be simpler and more basic based on the necessary background. Because for a beginner, they may be lost from the theoretical words and have difficulties understanding what you have down. It would be a better way to introduce some molecules containing the group you have calculated.
Author Response
Comment 1: Can you explain how the Duschinsky matrix facilitates the understanding of vibrational coupling between the ground and excited states? What insights does it provide regarding the vibrational modes in the S1-state? (page 7, paragraph 1)
Answer 1: The magnitude of the (3N-6)(3N-6) elements of the Duschinsky matrix show, which modes in the ground state resemble those of the excited state. Matrix elements with a value close to 1 signify that the respective mode in the S1 is comprised solely of the same mode in S0. This was added to the main text.
Question 2: What significance do the changes in the CN bond length have on the overall electronic and geometric properties of the chromophore? How might these changes affect the molecule's photophysical behavior? (page 11)
Answer 2: The changes in the CN bond length upon pi -pi* excitation follow the ring expansion. The strong decrease of the C-CN bond length shows a considerable stabilization of this bond and an overall stabilization with respect to dissociative photoionization channels, that are known for benzonitrile.
Question 3: How do the results from the SCS-CC2/cc-pVTZ level of theory compare with other computational methods (e.g., DFT, MP2)? Are there specific advantages or limitations of the SCS-CC2 method in predicting bond length changes? (page 11)
Answer 3: The main reason for using SCS-CC2 is that for this method the Hessian is available at the same level of theory for both states. This of course is also true for DFT/TD-DFT, which compare, however, not so well with the experimental vibrational frequencies. MP2 for the ground state is of comparable quality, but lacks a good S1 counterpart. The spin-component-scaled (SCS) version of CC2 has to be used, for aromatics that contain hetero atoms. Pristine CC2 has large deficiencies, even for the correct order of excited states, which are removed by spin-scaling. Other scaling methods like spin-only-scaling perform equally good than SCS.
Comment 4: What are the implications of the symmetric increase in aromatic ring CC bond lengths for the electronic and vibrational properties of 1,3-DCB? How might this affect its photochemical behavior?
Answer 4: Since 1,3-DCB is a moderately strong CH acid, it can be expected that electronic pi-pi* excitation leads to a stronger acidity, due to the possibility of stabilization of the formed anion due to increased pi resonance interaction. We will investigate this interesting aspect in a future publication.
Comment 5: For easier reading and understanding, the introduction and content should be simpler and more basic based on the necessary background. Because for a beginner, they may be lost from the theoretical words and have difficulties understanding what you have down. It would be a better way to introduce some molecules containing the group you have calculated.
Answer 5: We added the necessary background for understanding how the Franck-Condon analysis works into the supporting information, due to limited space in the manuscript. We hope this clarifies the problems.
Reviewer 4 Report
Comments and Suggestions for Authors
The works by Martini, Amar, and Schmitt present the application of a strategy (named FCfit) based on fitting the nuclear displacements along normal modes to maximize the similarity of the resulting vibronic fluorescence spectrum to the experimental one. In this way, it is possible to derive the geometry of the excited. The work is interesting, as it offers a direct way to obtain otherwise elusive structural information and merits publication in the Journal. However, in its current form, the paper lacks some information, which makes it confusing at some points.
Namely, the FCfit procedure should be included in more detail, including:
- The expression adopted to compute the spectra, clarifying the initial vibrational state depending on the exctitation. Does the temperature have any influence? Does the system have time to reorganize vibrational energy before emitting it?
- How do displacements in Cartesian coordinates enter the expression to be minimised?
- Is there a target function that is minimized? Does the minimisation include all spectra (i.e., for different excitation wavelengths) with the same weights?
- From the SI, it seems that one of the spectra is discarded from the fit. Could the authors explain why?
- From Fig. 2, it seems that the S0 geometry is taken from the experiment, as the method only provides displacements. Could the authors clarify this point
- How does the vibrational analysis performed at S0 and S1 affect the results (level or theory)? Are normal modes evaluated in Cartesian or valence internal coordinates?
- Regarding the calculated spectra, the authors could provide more details on the deviations using the Cartesian displacements from the optimized structures. Are they a consequence of errors in the S0 and/or S1 geometries? Could the normal mode analysis affect the results (using Cartesian or internal coordinates, or adopting vertical models, i.e. expanding the final, ground, state PES around the minimum of the initial, excited state?
Author Response
Comment 1: The expression adopted to compute the spectra, clarifying the initial vibrational state depending on the exctitation. Does the temperature have any influence? Does the system have time to reorganize vibrational energy before emitting it?
Answer 1: The excited vibronic state is also the emitting state, there is no IVR on the timescale of the experiment. Temperature does not play a role here, except for the fact that under the ultracold conditions of the molecular beam no collisional relaxation takes place.
Comment 2: How do displacements in Cartesian coordinates enter the expression to be minimised?
Answer 2: We apologize for the short description of the program in the manuscript, but were forced to omit a lot of details, that are crucial for understanding, due to similarities with prior papers. We decided to enter all this information in the supporting material. The displacements enter through the linear orthogonal transformation of the ground and excited state normal coordinates given by Duschinsky.
Comment 3: - Is there a target function that is minimized? Does the minimisation include all spectra (i.e., for different excitation wavelengths) with the same weights?
Answer 3: The target function is the weighted sum of the squared errors χ2 of the intensities of all spectral intensities in all spectra. We used these with equal weights, although our program allows for different weights (in case of noisy spectra, which should enter the evaluation with a lower weight).
Comment 4: From the SI, it seems that one of the spectra is discarded from the fit. Could the authors explain why?
Answer 4: We added the following phrase to the caption of Figure 4: An additional emission spectrum has been recorded, which we were not able to fit (Figure S6 of the online supporting material)
Comment 5: From Fig. 2, it seems that the S0 geometry is taken from the experiment, as the method only provides displacements. Could the authors clarify this point
Answer 5: Figure 2 shows only the calculated TDM orientation for excitation to the S1, in the inertial axis system of the ground state. For the FC analysis, the S0 geometry is taken from the experiment and distorted along selected normal coordinates.
Comment 6: How does the vibrational analysis performed at S0 and S1 affect the results (level or theory)? Are normal modes evaluated in Cartesian or valence internal coordinates?
Answer 6: The level of theory for the vibrational analysis affects the results strongly. We added to the Section Theoretical Methods the following: "The main reason for using SCS-CC2 is that for this method the Hessian is available at the same level of theory for both states. This, of course, is also true for DFT/TD-DFT, which, however, do not compare very well with the experimental vibrational frequencies. MP2 for the ground state is of comparable quality, but lacks a good S1 counterpart. The spin-component scaled (SCS) version of CC2 has to be used, for aromatics that contain heteroatoms. Pristine CC2 has large deficiencies, even for the correct order of excited states, which are removed by spin-scaling. Other scaling methods like spin-only-scaling (SOS) perform equally good than SCS."
Normal modes are calculated in Cartesian coordinates using numerical second derivatives of the potential energy.
Comment 7: Regarding the calculated spectra, the authors could provide more details on the deviations using the Cartesian displacements from the optimized structures. Are they a consequence of errors in the S0 and/or S1 geometries? Could the normal mode analysis affect the results (using Cartesian or internal coordinates, or adopting vertical models, i.e. expanding the final, ground, state PES around the minimum of the initial, excited state?
Answer 7: Very good point, indeed. We added to the Results section: "The vibronic emission spectrum, which is obtained by excitation of the vibrationless origin, is mainly governed by overtones and combination bands of the modes Q$_7$, Q$_{20}$, and Q$_{24}$. These modes are depicted in Figure S1 of the supplementary material available online. It is remarkable how the fit of the geometry changes improves the intensity features compared to the simulation, performed at the \textit{ab initio} geometry of the excited state. This is especially true for the overtone of mode Q$_{20}$ and combination bands of mode Q$_{20}$ with Q$_{24}$. Obviously, the \textit{ab initio} geometry is considerably distorted along these modes compared to the ground state in contrast to the experimental finding." and to the Discussion: "All ring CC bond lengths increase between 2.6 and 4.0 pm, which can be expected for a $\pi\pi^{\ast}$ excitation. The largest difference between \texit{ab initio} calculated and experimentally determined ring bond lengths is found for the C(4)-C(5) and C(5)-C(6) bonds. Here, the theoretical and measured bond lengths differ by 5 pm, the former being longer than the latter. It is this difference that can be compensated for by a combination of the modes Q$_{20}$ and Q$_{24}$. "
Round 2
Reviewer 3 Report
Comments and Suggestions for Authors
The revised paper can be accepted at the present form.