TP Computational chemistry: spectroscopy and excited states for luminol.

M1 SdM Chemistry.

Date: Oct. 2021

E. Dumont, S. Steinmann, P. Colinet

This practical aims at illustrating on a small size, yet interesting molecules the use of DFT to assign spectroscopy signatures, and also to show how it can impulse strategies for chemiluminescent molecules. If you never heard about luminol, you can have a glimpse here youtube link.

On this small-size system, we will showcase the power of DFT to generate several spectra... and see some pitfalls to be avoided or inherent limitations. It is inspired from the following paper: Giussani et al., Chem. Eur. J., 2019, 25, 5202–5213. A pdf version can be found on etudes.

  1. A first DFT optimization and generation of RMN signatures
  2. A bit of reactivity... with a twist !
  3. Mapping excited states surfaces

1. First DFT optimization and generation of RMN signatures

Objective: illustrate that the performance of a density functional is not decisive for standard geometry, but directly impacts on electronic properties

This time, one can build with no ambiguity the starting structure for luminol. You will perform a geometry optimization at the B3LYP/6-31G(d) level of theory, and ask Gaussian09 to also compute the NMR spectra for comparison with the experimental signature. The header to call a NMR calculations is given below:
#P nmr pbe1pbe 5D pop=reg IOP(6/7=3,6/17=2,6/26=4) gfinput 6-311G** scrf(solvent=dmso,pcm)


 0 1

(blank line) 

Up to you !

  1. Build the luminol molecule or grab a structure from Internet, and perform a geometry optimization at the B3LYP/6-31G(d) level of theory. Report the energy for further use.
  2. A spectroscopic signature of interest is the NMR one, for which experimental reference spectra are available. We can find them for instance on a Japanese databank website for spectroscopic properties of small molecules: database, lien vers le luminol. Perform a calculation at the PBE0/6-311G(d,p) level of theory, starting from the geonetry of luminol you optimized before. The chemical displacements correspond to the line "Isotropic=". To be able to compare with experimental spectra, you will need to perform a calculation for tetramethylsilane (TMS), at the same level of theory. Gaussian will give you only isotropic shielding constants: you can convert them into atomic chemical shifts by reference to tetramethylsilane (TMS) for which a NMR output file should be obtained at the same level of theory.
    $grep Isotropic luminol-nmr.log 
          1  C    Isotropic =    47.0322   Anisotropy =   175.3419
          2  C    Isotropic =    65.2666   Anisotropy =   174.7674
          3  C    Isotropic =    41.4138   Anisotropy =   201.9408
          4  C    Isotropic =    59.7629   Anisotropy =   149.6801
          5  C    Isotropic =    24.5913   Anisotropy =   185.2194
          6  C    Isotropic =    72.8334   Anisotropy =   144.8718
          7  H    Isotropic =    24.3009   Anisotropy =     6.0285
          8  H    Isotropic =    24.0036   Anisotropy =     4.6379
          9  H    Isotropic =    24.6712   Anisotropy =     6.1282
         10  N    Isotropic =   182.3838   Anisotropy =    72.3930
         11  C    Isotropic =    12.6761   Anisotropy =   126.2220
         12  N    Isotropic =   106.2568   Anisotropy =    73.2880
         13  O    Isotropic =   -18.6362   Anisotropy =   520.6330
         14  N    Isotropic =   103.3202   Anisotropy =    73.8927
         15  C    Isotropic =    16.7430   Anisotropy =   119.6854
         16  H    Isotropic =    25.3201   Anisotropy =     8.6123
         17  O    Isotropic =   -33.2320   Anisotropy =   568.9601
         18  H    Isotropic =    25.4569   Anisotropy =     8.2522
         19  H    Isotropic =    23.6383   Anisotropy =    14.9325
         20  H    Isotropic =    27.4631   Anisotropy =    11.5831
    Compare the predited values against the experimental assignment proposed for the 13C and 1H NMR spectra. What is the main difference between the two 1H NMR spectra reported experimentally ?
  3. We now would like to compare the results obtained with seven different level of theory, which are given at the end of this document in a archive. Which functional would we recommend ? To what extend do they differ ? Feel free to use (and to improve !) the python script given in the archive. Does the performance follow the Jacob's ladder ?
  4. Which other levels of theory could provide a reference set for benchmarking a functional ?

Beyond a single molecule, the performance is supposed to be similar along a family of molecules. This allows to define pre-calculated coefficients to enforce the comparison: cheshirenmr server

Remark: from DFT, it is also possible to extract spin couplings by specifying the ad hoc key word (e.g. luminol-nmr-constants.log) and gain more insights. To obtain a better numerical agreement, one would need to include dynamical effects and average NMR chemical shifts over time, accounting for molecular vibrations, explicit solvent, temperature...

3. A bit of reactivity... with a twist

Objective: illustrate the charge transfer... which can be a problem for DFT.

We will now be considering the reaction of luminol with oxygen, either in the triplet state or in the singlet state. The latter is not the ground state of molecular dioxygen, which lies normally as a triplet state. The mechanism given below implied anionic species, six-member rings with N-N and O-O bonds which implies a lot of dynamical electron correlation.

some_text Up to you !

  1. Propose energies situating the double deprotonation of the luminol. What is the effect of the solvation (we will consider DMSO, the keyword is scrf=solvent=dmso to call the IEFPCM mechanism) compared to the gas phase situation. The corresponding keyword to account implicitly for solvation is scrf=solvent=dmso.
  2. Where is the charge "located" for the dianion ? Can you trace back the main change from the structural point of view ?
  3. Lay down an energetic profile assuming the reaction takes place in the ground state. Calculations will be performed at the B3LYP/6-31+G(d,p) level of theory. The following intermediate is given as a hint, where the O-O distance is constrained: approach-o2-luminol-constrained.log. Why do we change the basis set from now on ?
  4. Plot the profile and conclude on the feasability of the reaction at the ground state. Which intermediates are difficult to situate ? Characterize the nature as intermediate of reaction or transition state of the bridge structure display below.

    It is difficult to reach the convergence the complex luminol with an interacting 1O2 molecule. Why ? We will approximate the energies of the complex as the sum of the energies of the separated fragments.

4. Mapping the triplet ground state profile into a singlet state excited one.

The chemistry of luminol takes place with a source of light... which may have a role !
  1. A first calculation of the excited state can be performed on the intermediate between the luminol and 3O2. Instead of the ground-state version of DFT, we will perform a vertical TDDFT calculation (we provide the header to be used). One output is here as an example: approach-o2-luminol-tddft.log
    # b3lyp 6-31+G* td(Nstates=40,50-50) gfinput iop(6/7=3) scrf=solvent=dmso
  2. What is the spin state for the first excited state ? At which energy is it lying ?
  3. Can you assign the molecular orbitals coming into play for the first transition ? The photochemical character of this reaction can be explained by a Walsh diagram (given in the research paper).
  4. The script allows to plot the final UV-Vis spectra after convolution, to mimick the broadening inherent to an experimental spectra (where close-lying transitions can merge into one peak).
    python approach-o2-luminol-tddft.log 
    Computation type ? (CIS, TD-DFT...) TD-DFT
    Enter number of computed States: 40 
  5. How would you proceed to go beyond the hypothesis of a vertical TDDFT calculation ?

Last update: Oct. 2019


Spectrum obtained for the approaching structure after python's "treatment"