Winter school – Exploring reactions using metadynamics

 Michaël addition

Date: January 2024

C. Michel

This session follows a first session dedicated to the Michael addition of acetone on the trans-nitrosytrene using static approaches with Gaussian. We then identified an intermediaire and a two-steps mechanism. Starting from the adduct, the retro-Michael starts with the C-Hsplit/O-Hbond formation and is followed by the CC split.

Our calculations were performed / will be performed at the PM6 level for the sake of efficiency using CP2K. The manual can be found here online. Input files (and some output files) and the protocol to submit a calculation will be provided during the practical session. The visualisation of results can be done using VMD, python and/or gnuplot.

The sequence is split in four steps and is expected to take 4h00, not including the simulation time. Working as a team is encouraged.


1. Molecular dynamics of the intermediate and the Michael adduct

    Prior to metadynamics, one has to analyze a molecular dynamics trajectory to select appropriate collective variables and parameters for the repulsive gaussians.

  1. In the directory Intermediate/PRELIMINARY, you will find the CP2K input file (PM6.inp) and some outputfiles for (i) the geometry optimization and (ii) a molecular dynamics starting from the optimized structure. Explore the files.
  2. Is the optimized structure identical to the one found using Gaussian?
  3. As which temperature the molecular dynamics has been performed? What is the thermostat? You can check the time evolution of the temperature and energetic quantities that are reported in the .ener file.
  4. Plot the time evolution of selected distances, angles and dihedral that you might considering as a collective variable and identify the degrees of freedom that are indeed free at the chosen temperature. This can be done uploading the trajectory file Intermediate_MD-pos-1.xyz using VMD.
  5. Repeat the same analysis on the Michael adduct. The corresponding files can be found in Product/PRELIMINARY
  6. Propose several set of 2 collective variables that could distinguish the Michael adduct, the intermediate and the products.


2. Exploring various set of collective variables

  1. In the directory Product/META_SizeFreqGaussian, you will find several directories. Each of them contains the files of a metadynamic run starting from the last Michael adduct structure of the molecular dynamics, using a set of two collective variables, the CC distance and the CH distance. Analyse the results of the directories starting from 01_xx. We detail here a protocol of analysis:
    1. Visualise the metatrajectory using VMD. Note your observations regarding the observed reactive events but also the other degree of freedom (for instance, rotation of the methyl group). If you observe any suprising structure, check how likely it is lauching a geometry optimization.
    2. Read the .inp file. Identify the collective variables. Check for the presence of walls.
    3. Identify the gaussian parameters (width, height, how often they were added).
    4. Plot the evolution of the colvars in function of time (COLVAR file) and overlap when the gaussians have been added (HILLS file). The CV should vary and the gaussian should not be added ‘on the top of each other’.
    5. Plot the localisation of the gaussians in the (CV1;CV2) plane (CV2 in function of CV1 using the HILLS file). This plot shows the chemical space that has been affected by the bias.
    6. Build the free energy surface using graph.psmp, the options -ndim 2 -ndw 1 2 -cp2k -file and the .restart file as an input. You should obtain the fes.dat file that contains the energy in Hartree in function of CV1 and CV2.
    7. Plot the free energy surface using fes.py.
    8. Using options of graph.psmp, you can also plot 1D profiles, but also 1D profiles at different accumulation time (with more and more added gaussians for instance). Analyse the evolution of the bias with the accumulation time.
  2. In 02_xx, 03_xx, etc. the same set of CVs is used, but using other parameters for the gaussians. Follow the same protocol of analysis. Conclude.
  3. In the directory Produit/META_ChoiceCV, you will find several directories employing different set of collective variables. For each set of CVs, two simulations are provided using different gaussian parameters.
    1. 01_distanceCC_distanceOH: Using the OH distance is an alternative to the CH distance.
    2. 02_distanceCC_distancesOH-CH: When considering the transfert of an atom H between two atoms, here C and O, using the difference between the two distances (CH and OH) allows to bias this transfer and not only the bond splitting.
    3. 03_cnCC_cnCH and 04_cnCC_cnOH : Using distances to distinguish a dissociated state from an associated one leads to a projection that is ‘unbalanced’. The volume to explore in the dissociated space is much bigger than the one corresponding to the non-dissociated case . The variation of the CV in the two wells is also contrasted (flat well vs. narrow well), making the choice of the gaussian parameters complicated (they cannot be nicely adapted to both situations). Using coordination numbers is a way to circumvent those issues. However, other issues may raise that you may observe analyzing those two runs.
    4. 05_distanceCC_cnCHminusOH_cnCO: Using three CVs offer more possibilities but complexify the visualization and requires a longer sampling (which scales in Nd with d the number of CVs).
  4. To improve the quality of the free energy surface obtained from the metadynamics bias, one can use a well-tempered metadynamics. Analyse the results in META_WT.


3. Starting from the intermediate

  1. Using your favorite set of variables, prepare an input file to launch a metadynamics to explore its transformation into the Michael adduct and/or the enol/transnitrobenzene. The preliminary runs can be found in the Intermediate directory.
  2. Launch the run and monitor it regularly. It may take few hours. Compare with the results obtained starting from the Michael adduct.


4. Without and with a catalyst…

  1. Using your favorite set of variables, prepare an input file to launch a metadynamics to explore its transformation into the Michael adduct and/or the enol/transnitrobenzene in presence of a catalyst. The preliminary runs can be found in the Intermediate directory.
  2. Launch the run and monitor it regularly. It may take few hours. Compare with the results obtained without a catalyst.